Chapter 6

Why Were No Anticancer Effects Detected?

Inspiring Others to Carry Out A Randomized Clinical Trial

In 1973 Pauling began his efforts to inspire other researchers to carry out a randomized double blind study of vitamin C in cancer patients. At this time he still had considerable influence in the research community. His prestige had been eroded, especially within medicine, by what were believed to be reckless claims for the value of vitamin C. Still, it seemed reasonable that he would be able to interest someone in doing the double blind study. This was finally achieved, but only after four years of effort.

As mentioned in the preceding chapter, in 1973 Pauling had first tried to interest oncologists at Stanford University, and found them completely unresponsive.

Pauling was turned down by the American Cancer Society. For a number of years he was also refused by the National Cancer Institute (NCI). On his first visit to NCI in 1973, Pauling showed the case histories of Cameron’s first 40 ascorbate treated patients to senior officials at the NCI. These officials could not be convinced that anything of interest was happening in Cameron’s patients. This seems rather astonishing. The patients who Cameron was treating were advanced cancer patients who were considered untreatable, and who were in poor condition. Although we can't know exactly the probability of spontaneous improvement or remission in such a group, it was clearly very low. A single remission would be extremely improbable, and a number of remissions would be extraordinary.

This kind of response by mainstream oncologists was to persist. Many years later, in 1989, Pauling met with the head of the NCI and described some of the remarkable remissions seen with ascorbate treatment. The head of the NCI set up a committee to review 25 case reports. Again it was concluded that there was nothing of interest in those case reports.

It has been suggested (Hoffer, 2001), that the oncologists reviewing these cases was not looking to see whether vitamin C had plausibly produced benefits, but instead were demanding ironclad proof . They would only consider the claims for vitamin C if every other possible cause of improvement could be rigorously excluded. As Hoffer notes, in considering whether a new therapy might be worth investigating, the plausibility of beneficial effects should be sufficient. It would be absurd to demand definitive proof at that preliminary stage of enquiry.

In 1973 Pauling was also told that a human trial could not be carried out until the safety and efficacy of the treatment had been demonstrated in animal studies. This was standard policy, reflecting the NCI’s rules for the evaluation of new cancer therapies. On the other hand, it could also be seen as a logical absurdity. The patients to be treated were at death’s door, and any danger from large doses of vitamin C was certainly minute compared to that from the cancer that was about to kill them. By this time, in addition to Cameron, many doctors out of the mainstream had clinical experience with high-dose vitamin C and could testify to its safety. This demand from the NCI illustrates the mechanical character of the mainstream war on cancer.

But in any case, it appeared that animals studies of safety and efficacy would have to be done before support could be found in the United States for a clinical study. When Pauling applied for a $100,000 grant to perform animal tests it was not approved. It was not until 1981, and many grant applications later, that Pauling was able to get support for this (Sun, 1981). One might argue that the granting process was overly conservative, and there should have been more appreciation of the potential of this novel non-toxic approach to cancer therapy. But it also appears that Pauling was perceived as a difficult character, and there were failures of Pauling and the Linus Pauling Institute to cross all the t’s and dot all the i’s in the application process.

In the meantime, before the animal studies were ever carried out, a more sympathetic NCI official (Vincent DeVita) approved sponsorship of a study of vitamin C effects in cancer patients, and allowed the clinical study to be carried out without prior animal studies. The person chosen to carry out the new study was an eminent cancer specialist, Charles Moertel, of the Mayo Clinic’s cancer center. Pauling communicated with Moertel about the need to use as subjects, patients who had not been exposed to chemotherapy or radiation treatment that might damage their immune systems. As we will see in the next section, this was not done.

The outcome of the Mayo Clinic’s first study, was entirely negative except for some suggestion of palliative effects. Pauling, by loudly raising objections to the conduct of that study, was able to force another, whose conclusions proved equally negative. In some ways, this new study was very different from Cameron’s original work, and in the eyes of Cameron and Pauling, a very imperfect replication. Once again, there was an angry public response from Pauling, but few were still willing to listen. From the beginning Cameron and Pauling had found it difficult to publish work on vitamin C and cancer in medical journals. Now it was almost impossible.

The First Mayo Clinic Study

The first of the Mayo Clinic studies, (Creagan et al., 1979), provided data on 123 terminal patients, "...unsuitable for treatment with systemic chemotherapy, either because of progression of disease after previous efforts or because their general condition precluded cytotoxic regimens." 60 of these received vitamin C and 63 placebo. Vitamin C (10g/day) was given in the form of five .5 gm capsules four times a day. Subjects were randomized in such a way that the groups did not differ with respect to cancer site, level of functioning, and grade of their cancer. Subjects received vitamin C until death or until they were no longer able to take medication by mouth.

Every two weeks a checklist was used to determine patient’s status with respect to four symptoms. The results are shown below.

That the denominators in these ratios are smaller than the numbers of patients in each group, evidently means that some of the patients' data was rejected for unknown reasons. It would be nice to know more details of how data about symptoms was collected.

This data presents patients in a binary way, as having either experienced or not experienced an improvement in a symptom during treatment. What kind of patient response resulted in a patient being scored as improved? Was a patient counted as having experienced improvement if improvement was indicated only on a single occasion during one two week period? What was the time course of improvements? Some presentation of the time course of symptom responses would have been helpful.

The percent of patients who reported improvement in any of the four symptoms was almost identical for treatment and placebo groups, 63% and 58% respectively. This difference was both non-significant statistically and apparently trivial in magnitude. However, reporting the difference in this way fails to express all the information about symptoms provided by their table. The percentage of patients who reported one or more symptomatic improvements was only a little higher for the ascorbate, than for the placebo group. But, evidently, the ascorbate patients who reported improvement on one symptom were also more likely to report improvement on other symptoms. If the percent of patients reporting improvement on any of the four measures is summed within the groups, the totals are 118 for the ascorbate group and 93 for the controls, an almost 27% difference. Apparently improvements were concentrated in certain individuals who must have showed improvements in multiple symptoms. The report of a mere 5 percentage point difference between groups is compatible with some treated subjects showing major improvements, and compatible with clinical reports of major palliative effects in treated patients.

While there were no significant differences in symptoms between the two groups this was barely the case. If we add up the number of positive responses in the numerator and the numbers of respondents in the denominators, we get a total proportion of 62/208 (.298) for the ascorbate group and of 48/206 (.233) for the placebo group. This proportion for the ascorbate group is almost 28% higher than for the controls. If we use a one tailed test for a difference between two proportions, this falls just short of significance at p=.06. The same test applied to strength gives a p value of .023. But if this value is adjusted for the fact that there were four different possible comparisons between groups, it too fails to reach significance.

However, the fact remains that the percentage of subjects reporting increased strength was twice as high in the ascorbate group than among controls, while percent reporting reduced pain was almost 60% higher in the ascorbate group than among controls. This is more in agreement with the evidence for these effects from earlier studies than it is with the null hypothesis advanced by Creagan et al.

On the other hand, the failure of the other two measures, appetite and activity level to be greater for the ascorbate group certainly is negative evidence. Appetite is especially significant because improvements in appetite were frequently reported in earlier studies. Increased activity level was not as clearly expected from the earlier reports. Perhaps effects on strength and pain are of the greatest interest because these are frequently reported in the clinical literature and because they seem to have a respectable biochemical rationale. These may be mediated through ascorbate’s effects on carnitine metabolism and excretion and ascorbate effects on prostaglandin metabolism. Studies concerned with these effects and the mechanisms which might mediate them are described in chapters seven and eight.

In remarks at a workshop on vitamin C, Pauling (1982), stated that the differences in palliative effects were statistically significant. However, he seems to have been largely wrong about this. It was true that the effect on strength would have been statistically significant in isolation, but only by picking out this comparison which showed the expected direction, and ignoring the others which were not. Pauling also said that the difference in pain symptoms was significant, which as far as I can determine is not true.

In Cameron's clinical reports that did find benefits of ascorbate, only some patients showed these symptomatic improvements, and in many cases for relatively brief periods of time. This period of time usually began about 5 to 10 days after the beginning of ascorbate administration. In this Mayo Clinic study the symptom checklist was filled out every two weeks through the duration of the study. Because of the way in which symptoms were registered (improvement or no improvement) patients in whom powerful benefits occurred could not be distinguished from those in whom marginal effects were seen. Information about degree of improvements was thrown away. Such information could have been obtained by using rating scales.

It would seem that small and temporary effects (persisting for a week or two) could easily be missed if they were only monitored once every two weeks. If there were any kind of averaging used over the course of the mean 50 day survival of these patients this would also make it more likely that effects could be missed. With these often transient effects appearing in only some patients, it might be necessary to collect much more data than was done in order to solidly establish their existence. Because this was a randomized double blind study (it wasn't know who was receiving ascorbate) there was no close observation of individual patients to note the course of their response to the treatment. In this and the other Mayo Clinic studies there was no close clinical observation of patients like that at the Vale of Leven.

There was no suggestion of improved survival with ascorbate. Survival curves for treated and control groups were almost identical, but the last two, relatively long lived survivors, were from the placebo group. The mean survival period from the beginning of the study was 51 days for all subjects, while in the original analysis of the Vale of Leven results (Cameron and Pauling, 1976) it was 50.4 days for controls. This population seems to have rather exactly matched the original Vale of Leven control group in terms of how advanced the patient’s illness was.

Pauling had suggested that patients who had received chemotherapy might suffer from immune suppression and therefore be less suitable candidates for ascorbate treatment than Cameron's subjects. His patients had rarely received chemotherapy. Creagan et al acknowledged that, "...patients with advanced cancer who have previously been treated with irradiation or chemotherapy are indeed immunosuppressed". However, they cite data from two samples of cancer patients who they believed resembled those in their study, and who still showed considerable immune response. In those two samples, 80% in one case, and 65% in the other had responded to skin tests thought to indicate immune competence relevant to the destruction of cancer cells. No doubt these data provide some kind of evidence for the Mayo Clinic's point of view.

The Second Mayo Clinic Study

Another study was carried out, designed to meet the objection of Cameron and Pauling, that ascorbate might not work in subjects who had extensive exposure to chemotherapy.

A Third Mayo Clinic Study

Finally, between the two studies described above, there was a third study carried out jointly by the Mayo clinic and a local cancer center. This study was only described in an abstract (Tschetter et al., 1983). This was another double-blind study which employed the same forms and dosage of ascorbate as the other Mayo Clinic studies. The subject population was composed of individuals in whom conventional treatments had either failed or were "inappropriate". In this study, patients were maintained on ascorbate as long as they lived or as long as they were able to take the medicine orally. It was reported that in the ascorbate group there was a non-significant trend toward improved appetite, strength, and pain control. Again there was no effect of ascorbate on survival.

Why Did the Mayo Clinic Studies Not Confirm the Reports of Cameron and Pauling?

In explaining the different outcomes at the Vale of Leven and at the Mayo Clinic, we will consider in detail many features of the two studies. However before doing this we will review some of the background evidence for the reality of ascorbate effects against cancer. If this evidence is strong enough, it justifies an attempt to explain the different outcomes of the studies on the basis of something other than Cameron et al's methodological incompetence.

Many studies have looked in vitro or in experimental animals, for anticancer effects of ascorbate, of ascorbate analogs, or of ascorbate in combination with other agents. These studies are reviewed in chapter 7. Although some of these reports are negative, and a few seem to convincingly indicate that ascorbate can exert procarcinogenic effects, they show a clear predominance of anticancer effects.

Some of these positive studies cannot be regarded as directly supporting the findings at the Vale of Leven. This is because these studies have employed much higher doses than were used in the human studies, because they used ascorbate analogs, or because they used ascorbate in combination with other agents. Nevertheless, almost all of these studies suggest that there are real anticancer effects associated with ascorbate. When considering the general plausibility of the anticancer effects reported by Cameron and denied by the Mayo Clinic workers, it is important to keep in mind this background of established anticancer effects in vitro and in experimental animals.

The Two Independent Discoveries of Ascorbate’s Palliative Effects in Cancer Patients

The preceding chapter described how the palliative effects reported by Cameron had also been reported a generation earlier in German language medical journals. The similarity of these early reports to those of Cameron and his associates is striking. There have been two entirely independent discoveries of these effects. Evidently Cameron had no knowledge of these early reports when he began his work with vitamin C.

That these two independent discoveries represent identical delusions, springing up independently and in the absence of real effects, seems enormously improbable.

An Existence Proof of Ascorbate’s Therapeutic Effects

In the previous chapter an enormously compelling case history was cited (Cameron et al, 1975, 1991). A patient was suffering from an advanced and rapidly progressing reticulum cell sarcoma. The patient experienced a prompt and complete remission when treated with ascorbate. This patient relapsed several months after ascorbate treatment was withdrawn. Resuming ascorbate treatment produced a second remission, though this time, it occurred more slowly and larger doses were required. This interpretation of what occurred was supported by a pathologist’s report on an excised lymph node, and by a complete X-ray record through the course of his illness. This material was repeatedly examined by outside experts.

Such patient’s are not known to spontaneously recover. The odds against a single spontaneous remission coincident with the initiation of ascorbate therapy are astronomical. The odds against two separate remissions occurring by chance, coincident with two ascorbate treatments, and in the absence of a real anticancer effects are even more astronomical.

This case was considered so compelling that when Cameron and Pauling presented this at Sloan-Kettering Memorial Hospital, their initially skeptical audience was won over.

If this patient’s history cannot be explained except by a genuine anticancer effect of ascorbate, then the failure of an anticancer effect to appear in three Mayo Clinic studies does not erase the significance of this case history.

Finally, there were a number of cases of tumor regression observed by Cameron and his associates among the other patients treated at the Vale of Leven. These cases, though not as exhaustively documented as the case described above, provide a similar kind of evidence, and if we are to take this case seriously, this provides reason to take the other cases seriously.

From the point of view of mainstream medicine, the obvious explanation for the differing outcomes was the methodological soundness of the Mayo Clinic studies, and the deficiencies of the Vale of Leven studies. These deficiencies would make it possible for the advocates of vitamin C therapy to translate their preconceptions, their delusions about vitamin C, into clinical observations of striking beneficial effects, and into numbers showing enhanced survival in their statistical analyses.

This point of view is developed most fully in the discussion section of the second Mayo Clinic study (Moertel et al, 1985). Cameron and Pauling were said to have "selected and reselected the patients they decided to evaluate". They compared their treated patients with historical controls. Cameron and Pauling had no protection against bias in the selection of their subjects. Therefore Moertel et al were "...left with the inescapable conclusion that the apparent positive results of Cameron and Pauling were the product of case-selection bias rather than treatment effectiveness." This conclusion would indeed be "inescapable", if the only meaningful differences between the circumstances of their studies and those of Cameron were methodological ones, and if the full range of information about ascorbate and cancer were neglected.

Suppose a medical researcher were presented with Cameron and Pauling’s 1978 paper describing the quantitative results at the Vale of Leven and with the papers describing the two published Mayo Clinic studies. The researcher, who probably starts with an initial skepticism about the therapeutic powers of vitamin C would unhesitatingly side with the Mayo Clinic and against Cameron and Pauling. After all, it is clear which side has methodological superiority. The methods used by Cameron and Pauling do not make it possible to rigorously exclude the possibility that bias might have entered into their selection of subjects.

Because at the Vale of Leven there was never a rigorously defined procedure (such as randomized assignment) to determine who went into treatment and control groups, the patients with superior prospects for survival might have been preferentially selected for treatment. Perhaps the data describing the groups do not suggest such a thing, or indeed, suggest the opposite. But still it remains a possibility. It is difficult to prove the negative, that there was no selection bias at the Vale of Leven. If it is accepted that the Mayo Clinic studies proved that ascorbate does not benefit cancer patients, then in some unspecified way, biased assignment to groups is almost required to explain the positive reports.

In a similar way, if Cameron and his coworkers, in their clinical reports described surprising benefits to their patients when they received ascorbate, it will be pointed out that their evidence is merely subjective, unsystematic, and anecdotal. The Mayo Clinic studies have proved that ascorbate does not benefit cancer patients. Therefore, if Cameron et al. believed they saw such things, they must have been deluded. Just how they were deluded is not important, but it must be assumed that they were somehow misled by their enthusiasm.

Surely it is perfectly sensible to consider methodological quality when evaluating data from conflicting studies. In general, and all else being equal, randomized double blind studies do provide more secure evidence than clinician’s impressions. Those who read the Mayo Clinic studies and knew nothing about the effects of ascorbate except what they found in Cameron and Pauling (1976, 1978) would have no reason not to prefer the Mayo Clinic’s numbers. After all, the Mayo Clinic had the superior methodology. On the other hand, as I will try to demonstrate, the totality of published information about the Vale of Leven findings makes a compelling case, that real ascorbate effects were demonstrated there. This is no less true for the failure of those same effects to appear in the studies carried out at the Mayo Clinic.

It seems to me that all the eminent medical men who dismissed Cameron and Pauling’s results were able to do so because they never thought about this totality of published evidence, and perhaps never read much of the relevant literature. Typically only a handful of studies are cited when dismissing the anticancer effects of vitamin C. The dismissal of the vitamin C evidence by medical conservatives also exemplifies a mechanical quality in their thinking. If the Vale of Leven studies were methodologically incorrect and the Mayo Clinic studies were methodologically correct, a decision can be rendered. There is no need to do any thinking about the particulars of those studies.

When should we take seriously a study which has methodological imperfections? When those methodological imperfections cannot explain the outcome, cannot realistically account for the shape of the data. This as I will try to show below is the case with the Vale of Leven studies. The size of the reported effect and the remarkable character of the effect described in the clinical report make it extremely implausible that these could be attributed to the biases that double-blind studies are supposed to eliminate.

It should be recognized that there were many differences between the contradictory studies besides prior exposure to cytotoxic therapies and the differences in methodology that riveted the attention of the Mayo Clinic workers. It is not clear, which, if any of these differences, might be important in creating differences in outcome. In fact, Cameron and Pauling seem to have shared the general presumption that the differences described below were not important.

The patient samples in one case were Scottish and in the other American. In the American case, they had generally received chemotherapy as well as surgery and radiation. While this was not true of the patients in the second Mayo Clinic study, that group differed from other study populations in being in markedly better general condition, and in the fact that they received ascorbate for a relatively brief period, relatively early in their illness. The Scottish sample (though there is no really precise information about this) seems to have distinctly poorer nutrition and perhaps more exposure to environmental carcinogens (Cameron, 1991). There is some suggestion that the Scottish subjects were from a lower socioeconomic group.

It is also true that Cameron’s patients received ascorbate in a different form than the patients at the Mayo Clinic. While the Mayo Clinic patients received sodium ascorbate in the form of 20 capsules a day, Cameron’s patients received ascorbate in the form of a sweetened solution which contained a good deal of sorbitol. It is possible that this solution resulted in better absorption than the capsules. Ascorbate solutions can become oxidized and produce breakdown products which have anticancer effects beyond those of ascorbate itself. (This is discussed in chapters 2, 7 and 8.) Many of Cameron’s patients also began their ascorbate treatment with approximately ten days of intravenous administration.

The role of these differences certainly seem worth thinking about. Each is discussed at greater length in one of the sections below.

The first Mayo Clinic study was criticized by Pauling and Cameron because the study subjects had prior exposure to chemotherapy, and this might impair their immune response. Very few of the patients originally studied at the Vale of Leven had been exposed to chemotherapy and it appeared that this might be a critical difference between the two study populations.

Since some part of the benefits associated with taking vitamin C were assumed to derive from immune system stimulation, this might make this group less responsive. This was the main reason why Pauling agitated for a second study. The conclusions of Moertel et al were that prior exposure to chemotherapy had been completely eliminated as a possible factor in the second Mayo Clinic study. There ascorbate had entirely failed to produce survival benefits in patients who had never received chemotherapy.

To Cameron and Pauling this was exasperating nonsense. By their failure to continue treating their patients with ascorbate as their illness progressed, Moertel et al ensured that they had not performed a valid replication of the original Vale of Leven study. The relatively short exposure to ascorbate is associated with the fact that this was a much longer surviving population of patients than those at the Vale of Leven or those in the other Mayo Clinic studies. The possible importance of this discontinuation of ascorbate is discussed at length in the next section.

According to Richards (1991) the Mayo Clinic workers had quite a different perspective regarding immune stimulation. They were thinking of vitamin C precisely as an aid for impaired immune systems. Patients whose immune systems had been ravaged by advanced cancer and cytotoxic therapies might be those most likely to be helped. It is interesting to think about this in connection with the chart that appears on page 125 of Cameron and Pauling (1993). This shows that cancer patients have lower leukocyte ascorbate levels than normals. Values are still lower in cancer patients who have received surgery, lower yet when radiotherapy is added to surgery, and lower still when chemotherapy is added to the other treatments. From a certain point of view, it would seem quite natural that this most treated group, with the most severe deficiencies, that supplementation should have the greatest benefits.

Among the non-conventional agents that have been used to treat cancer, a number are supposed to be immunostimulants. It might be of interest to see whether such agents have been reported to work in late stage cancer patients who have received extensive treatments. Examples of such agents would be extracts from maitaki mushrooms, beta glucan, vegetable sterols, cimetidine, etc.

As was mentioned above, Creagan et al (1979) cited data suggesting that patients like those in the first Mayo Clinic study retained a substantial level of immune responsiveness. This remains unsatisfactory evidence because there is no precise knowledge of what those test responses mean for resistance to cancer.

Another case in which successful ascorbate treatment was associated with the absence of prior cytotoxic therapy was the second Vale of Leven study (Cameron and Campbell, 1991). The patients there seem to have been quite similar to those in the original Vale of Leven study, though less sick, as indicated by a longer survival times for untreated controls.

There were the very terminal patients treated by Dr. Young at Sloan-Kettering Memorial (Richards, 1991, page 102) who had presumably already been extensively treated with chemo or radiation in the American fashion. No benefits of ascorbate were seen in these patients.

In a retrospective study of advanced cancer patients who were successfully treated with vitamin C plus other nutrients (Hoffer and Pauling, 1990), 85% of both controls and treated patients had received "...conventional therapy (surgery, radiation, chemotherapy)". Such treatments were certainly not an obstacle to the effects of ascorbate there. It is true that this treatment differed from that at the Vale of Leven and at the Mayo Clinic in that large doses of vitamin C were only part of the therapy. Nonetheless, this does not tend to support the explanation of outcome differences in terms of prior conventional cancer therapy. This study is described in more detail in the preceding chapter.

It does seem surprising, given the magnitude of effects reported in the original Vale of Leven study, that no ascorbate benefit on survival would have appeared in either of the Mayo Clinic studies. This would be true even if their patients were suffering from some reduction in immune responsiveness in the first Mayo study, and only received a truncated ascorbate treatment in the second. While it seems difficult to construct any kind of rigorous argument about this, it seems that prior exposure to cytotoxic therapy is not a very convincing explanation for the negative results in the first Mayo Clinic study. Other explanatory factors must be invoked.

Brief Duration of Ascorbate Treatment in the Second Mayo Clinic Study and Early Stage of Patients

As noted in the previous chapter, there didn't seem to be any clear distinction between cases which showed palliative improvement and those who showed increased survival. These patients seemed to exist on a single continuum, from minor symptomatic relief through extended survival. It might have been argued that palliative effects should be greatest specifically in debilitated late stage cancer patients, because they would have the most severe ascorbate deficiencies. At this stage massive supplementation could make up these deficits.

The colorectal cancer patients in the second Mayo Clinic study differed dramatically from the patients treated by Cameron et al. These patients were less advanced and were in much better condition. Most were without cancer symptoms when they entered the study. They were destined to live much longer from the study’s onset than Cameron’s patients had done. In the first Vale of Leven study, treatment was started at a point when colon cancer controls had an average time to live of 33 days and their rectal cancer controls had an average survival time of 43 days.

About 50 percent of the subjects in the second Mayo Clinic study lived for more than a year after the beginning of the study. Scarcely any ascorbate treated subjects died during the first ninety days, and by that time, only about 50% were still receiving ascorbate. By 12 months when about 50% of the treated patients were still alive, less than 10% were still receiving ascorbate.

We can summarize the situation by saying that the subjects in the second Mayo study received ascorbate for a relatively brief period early in their illness, while the subjects in the Vale of Leven study began ascorbate late in their illness, when severely ill, and continued ascorbate until their deaths. At the time the subjects in the second Mayo study began receiving ascorbate, they were evidently far less sick than Cameron’s patients when they began treatment.

It is interesting to think that in patients who received ascorbate for months, while their disease continued to progress, cell lines in their cancers would be selected for their resistance to the anticancer effects of ascorbate. As is discussed in chapters seven and eight, there is some reason to think that cancers may thrive on extra ascorbate at levels which are too low to exert anticancer effects.

There are also differences in the way ascorbate was administered. The initial administration ascorbate to many of the Vale of Leven patients was intravenous. While this procedure would produce much higher ascorbate plasma levels than oral ascorbate, this does not seem to be sufficient to explain the differences in outcomes. Seventeen of fifty subjects described by Cameron and Campbell (1974), only received oral ascorbate. Some of the dramatic effects reported in the Vale of Leven studies were seen in this group, including tumor regressions and tumor necrosis and hemorrhage.

While the subjects in all of the Mayo Clinic studies received 5, 500mg ascorbate capsules four times a day, the Vale of Leven subjects, four times a day, took a tablespoon of an ascorbate solution which also contained 23% sorbitol. Sorbitol is a sweetener added to many medicines, and which according to the Merck Drug Manual (13^th Edition), is used "to increase absorption of vitamins and other nutrients in pharmaceutical preparations". At the time when these studies were being carried out, no one seems to have seriously considered the idea that the form in which ascorbate was administered could be important. It is clear that initially, Cameron and his associates did not think so, "...the intravenous regime is probably unnecessary as a routine measure, and need only be employed in clinical situations, where vomiting, anorexia, or other complications of malignancy preclude oral administration." (Cameron and Campbell, 1974, p. 297). It doesn’t seem that Cameron or Pauling ever suggested that the administration of capsules rather than their ascorbate solution could make a difference. No evidence was ever gathered about the plasma levels produced by ingestion of sodium ascorbate capsules versus that produced by the solution used at the Vale of Leven.

It does seem possible that taking a spoonful of this sweetened solution four times a day might have been much more acceptable than taking the 5 capsules four times a day and led to better compliance.

It can be asked why on general principles, the Mayo Clinic workers should not have scrupulously tried to replicate the treatment used by Cameron and Pauling. Why not administer ascorbate in exactly the same way that Cameron had done. While internal evidence from Cameron and Campbell (1974) indicated that the difference between oral and iv administration was not important, the ascorbate solution Cameron used could have been duplicated with minimal effort.

There is also the possibility that an ascorbate solution might contain significant levels of ascorbate breakdown products that would be absent in ascorbate capsules. The most obvious possibility is DHA (dehydroascorbate) which will be present in oxidized ascorbate solutions. There are also other breakdown products which have been shown to possess anticancer effects beyond those of ascorbate (Yamafuji et al, 1971;Poydock et al, 1982, 1985;Tsao, 1991).

There is a possible role for threonic acid, another breakdown product of ascorbate. There is a commercial product, Ester C ©, described in a popular work, (Bland, 1995), which is an incompletely defined complex of ascorbate and calcium. This product has been found to be absorbed more rapidly, produce higher plasma levels, and appear later in the urine, than ascorbic acid (Bush and Verlangieri, 1987). It was also shown that a given quantity of ascorbate incorporated into Ester C ©, displayed considerably greater anti-scorbutic power than an equivalent amount of pure ascorbate (Verlangieri et al, 1991). The authors of these studies have referred to the presence of threonic acid in Ester C © as a possible basis for its increased activity. They have also published a study in which threonate was shown to considerably increase uptake of ascorbate into a T cell lymphoma line (Fay and Verlangieri, 1991). The obvious problem with this study and its conclusions is that these effects were found with relatively huge concentrations of threonate. The minimum concentration tested was 100 mg/dl. This seems absurdly irrelevant in assessing the effects of a compound which only composes 1.3% of Ester C ©. It seems much more probable that the increased effects of Ester C © relative to ascorbate is due to the form in which ascorbate is complexed with calcium, rather than to the presence of small amounts of threonate.

Block and Schwartz (1994) reviewing the literature about anticancer effects of ascorbate in animals and in cell culture, suggested that such effects were more frequently seen when ascorbate was administered in an animal’s drinking water than when provided in dry food. This would be because of the oxidation and breakdown of ascorbate that would occur in solution but not when ascorbate is incorporated into dry chow.

On the other hand, a description of the way in which ascorbate was formulated and stored at the Vale of Leven does not provide a great deal of support for this role of ascorbate breakdown products. The sodium ascorbate solution was relatively concentrated, it was stored in light-proof bottles and it was normally kept refrigerated. All of these factors should tend to enhance stability. Unfortunately, no one ever collected evidence about levels of breakdown products in the ascorbate provided at the Vale of Leven, nor about the effective plasma ascorbate levels produced in the Vale of Leven patients as opposed to those at the Mayo Clinic.

As was mentioned in the previous chapter, in describing the final Vale of Leven study (Cameron and Campbell, 1991), the authors stated that "...there seems to be a clear linear relationship between the highest recorded plasma ascorbate concentration and overall survival time, values of >3mg/dl being particularly desirable.". This is one of the few bits of concrete information about the relationship between plasma level and anticancer effects in the Vale of Leven work. It suggests that factors that modify absorption could be critical in determining whether an anticancer effect will be seen.

Among the letters to the NEJM following the publication of the first Mayo Clinic study was one which noted that in terminal cancer patients, secretion of digestive juices is reduced and absorption is impaired (Zaeslein, 1980). It was also noted that in such patients, the absorption of 5-fluorouracil from capsules was more variable than its absorption from liquids. This might explain why the effects of ascorbate from capsules might be different from that from a solution. However, this presumably would not be a significant factor in the second Mayo Clinic study where the patients studied were certainly not terminal in the same sense as in the other trials.

Zaeslein also pointed out that Cameron’s patients received their ascorbate together with sorbitol, and that sorbitol is used to increase the absorption of vitamins. He concludes by noting that blood levels were not checked to confirm absorption at the Mayo Clinic. In their reply, the Mayo Clinic authors (Creagan and Moertel, 1980), dismissed the possible role of sorbitol by simply saying that: "It is highly doubtful that the addition of a small amount of sorbitol would have any meaningful benefit in this setting."

I have looked a little into the literature on sorbitol and absorption. As Zaeslein noted, the Merc Index (Annonymous, 2001), indicates that sorbitol is used to increase the absorbtion of vitamins. The evidence that such a sorbitol effect was important at the Vale of Leven is hardly overwhelming, but this possibility is not excluded. The ascorbate solution used contained 23% sorbitol, and a tablespoon of this was taken (often after meals). This would then be diluted out by the stomache contents. It is true as the Mayo Clinic authors pointed out, that the amounts of sorbitol involved would be rather small. Most of the effects reported were observed with relatively high concentrations of sorbitol (e.g. 30%), though some striking effects on absorption have been described with a four percent solution (Garnier and Blanc, 1963). This study compared the uptake of sorbitol or glucose solutions, and of an equimolar solution which was half sorbitol and half glucose. The mixture produced absorption which at certain concentrations was substantially larger. If the presence of sorbitol could create a five fold difference in absorption at 4% concentration, perhaps it is possible that a tablespoon of 23% sorbitol solution could still exert some significant effect when diluted out below 4%. Of course appealing to such evidence also requires the assumption that findings results with sorbitol and glucose might have some relevance to the case of sorbitol and ascorbate. Unfortunately, ther simply doesn't seem to be any published evidence about the later.

This and other correspondence about the first Mayo Clinic study (four letters and a reply by Creagan and Moertel), appeared on pages 298 thru 300, vol 302 of NEJM.

What exactly was the rationale for taking patients off ascorbate or placebo when they were "progressing"? Was ascorbate supposed to be causing the progression? At the time when progression was detected, the authors would not know whether a patient was on ascorbate or placebo. It would not have been possible to know at that time that ascorbate was causing patients to progress. This only could have been determined after completion of he study. What if it was found at the end of the study that ascorbate treated patients were progressing more slowly than controls? In that case, the patients would have been harmed by terminating treatment. It is also true, that there was no effective alternative treatment for them to receive. Moertel himself had condemned the use of 5-fluorouracil in colon cancer as ineffective and a very unpleasant experience for patients.

While there were a few animal studies which suggested that ascorbate could exert procarcinogenic effects, the literature on human patients provided almost no suggestion of such an effect. The report of Szenes (1942) being perhaps the sole exception. Cameron and Campbell (1974) had noted that there seemed to be a remarkably accelerated progression in some of their ascorbate treated patients, but this was only after they had shown a period of clinical improvement and a halt or retardation in progression. But since Moertel et al thought that the effects claimed by Cameron and his associates were imaginary, it is hard to see why they would have paid attention to their account of such cases.

It seems unfortunate that the authors never published data showing how patients in each group progressed. If the ascorbate group was progressing faster, was it because they were losing weight faster, because their performance scores were decreasing faster, because they were developing new metastases, or because they were showing an increased cross section of their tumors? If ascorbate can in fact accelerate the progression of disease in cancer patients, this is certainly an important phenomenon. After all, at this time, many cancer patients were taking vitamin C. It would not be too much to ask that the researchers provide some information about the nature of this acceleration. Pauling and Herman (1989) in their attack on this study, mention in passing, that 19 of the ascorbate treated patients had an increased cross section of their tumors, while the remaining 32 showed some othe sign of progression.

If more rapid progression was due to increases in tumor cross section, could this reflect swelling at the site of their tumors as was reported by Campbell and Jack (1979) in their ascorbate treated patients. It was mentioned in the original clinical report by Cameron and Campbell (1974), that ascorbate treated patients occasionally reported pain at the site of metastases. This could have been a response to swelling at those sites. And this in turn might have been a manifestation of the tumor necrosis response described by these same authors. But in this case, this swelling would actually have been a manifestation of a beneficial action against the tumor.

There does seem to be a real possibility that the ascorbate was causing patients to do worse in the 1985 Mayo Clinic study. Although survival curves for the two groups are almost overlapping during the first year of the study, thereafter survival for the ascorbate patients is well below that of the controls. Inspection of the graph in Moertel et al. (1985) shows that around 19 months (the time of greatest divergence between the groups), only about 8% of the ascorbate patients were still alive and about 35% of controls. This result is puzzling because by 12 months only a very small proportion of the treatment group was still receiving ascorbate, and at this time, survival rates were still almost identical for the two groups. This seems to imply some kind of delayed effect of ascorbate exposure, like the rebound effect claimed by Cameron and Pauling.

It is perhaps surprising that the Mayo Clinic workers did not choose to stress the poorer final outcomes for ascorbate patients. They might have been expected to denounce vitamin C as a killer of patients and not merely ineffectual.

Cameron and Pauling believed that experience at the Vale of Leven, had shown that death often followed discontinuation of ascorbate in cancer patients. This at least was their impression. It does not appear that any objective evidence was collected about this point. There is however, evidence that adaptation occurs in human subjects receiving large doses of ascorbate. This was briefly discussed in chapter 1, and will be considered at more length in chapter 8. When subjects receive several grams of ascorbate per day, plasma levels may go back to baseline levels in a period between two or three and ten days. When the supplemental ascorbate is discontinued, plasma levels may fall below baseline and require a similar period to return to baseline. A study in which 10 g/day were administered reported similar effects (Tsao and Salimi, 1984). This means that after adaptation to high ascorbate intake, abrupt cutoff of ascorbate should result in subnormal levels in the body, what Cameron and Pauling referred to as the "rebound effect". If high intakes of ascorbate were playing an essential role in keeping cancer patients alive, then producing subnormal levels in the body by stopping ascorbate could reasonable be expected to put such patients in danger.

If the rebound effect was genuinely important in determining the survival of cancer patients, this might explain the lack of benefit from ascorbate seen in the second Mayo Clinic study. It would explain why there was a suggestion that the ascorbate treated patients had done more poorly than controls in terms of survival. They would have been harmed by the abrupt withdrawal of supplemental ascorbate and the resulting fall in ascorbate below unsupplemented levels.

After the second Mayo Clinic study, relations between Cameron and Pauling (Pauling in particular) and the Mayo Clinic workers, were very unfriendly. Because of this, Cameron and Pauling were not given access to the raw data of the second Mayo Clinic study. Therefore, they could only proceed on the basis of what they saw in the published report. They could only inspect the graphs in the report and make assumptions about what was happening to individual subjects. As we will see below, this apparently led them into some errors of interpretation.

Earlier, in describing the analysis of Vale of Leven data (Cameron and Pauling, 1976), and the analysis of some data collected by Dr. Hoffer, (Hoffer and Pauling, 1990), a technique was mentioned for describing the survival of cancer patients utilizing the Hardin Jones principle. This principle says that in a homogenous group of terminal cancer patients, a fixed proportion will die in each successive time interval. When survival is plotted against time, with percent survival on a logarithmic axis, and time on a linear scale, a straight line results. This generalization was based on the examination of many populations of terminal cancer patients.

If the population is homogenous, does not consist of subgroups of patients who have different survival prospects, then the line will be straight until the time when no patients survive. On the other hand, if it contains a subgroup which has a lower death rate, then after most of the high death rate subjects are dead, the line will bend upward and show a flatter slope characteristic of the low death rate subgroup. Similarly, if an effective treatment is instituted, the downward trending line for the treated group will turn upwards. If the treated and control groups are initially equivalent and the treatment has no effect, the graph of log percent survival against time should be the same straight line for both groups. On the other hand, only if something is being done, that actually harms patients, should the mortality slopes show a downward inflection.

Pauling and Herman (1989) looked at the data from the second Mayo study, using the Hardin Jones analysis. Pauling and Herman noted that in the period when most of the vitamin C treated patients were taken off of vitamin C, (70 to 120 days into the study) the death rate for the vitamin C treated patients was higher than at any time during the following 300 days. During this period, 10 vitamin C treated patients died, and only four controls. "The fact that this increased death rate occurred immediately after the median withdrawal date of the vitamin C, 75 days, suggests that it may have been caused by the rebound effect."

Pauling and Herman’s plot of log of survival percentage also shows the slopes for ascorbate and placebo treated subjects diverging sharply around 400 days after study onset. The line for the ascorbate treated subject turns abruptly downward, and then at about 550 days turns more sharply downward, and almost all ascorbate survivors are soon dead. At about 650 days, the curve for the controls seems to show a similar plunge. These downturns in the survival plot are taken to indicate that some treatment condition had changed for the subjects. This was interpreted as due to the participation of study subjects in conventional chemotherapy after closure of the ascorbate study. The fact that the lines for ascorbate and control subjects diverge also suggests that they did not represent equivalent groups and that either they differed initially, or that ascorbate administration had some effect.

The curves for the ascorbate and placebo groups do seem to deviate markedly toward the end, with the ascorbate group showing a distinct downward inflection. However, there are a number of problems with this analysis. These are discussed in Richards (1991, pp. 163-4). There was only a small number of data points in the graph for the periods where Pauling and Herman thought they saw important differences between the groups. In fact the differences lacked statistical significance. While the greater mortality in the ascorbate group at the end of the trial could be attributed to the discontinuation of treatment, it could equally logically be argued (as did the Mayo Clinic workers), that increased mortality was a delayed effect of ascorbate administration. Ascorbate had after all, been associated with more rapid progression.

Pauling also insisted that there must be something wrong with the data from the second Mayo study, because there was a lag period before the plot of the log of survival % began its downward slope. He noted that when data from hundreds of cancer studies had been examined, the data always showed a fixed mortality rate from the beginning as long as a homogenous population of patients was being studied. But it seems there is a very simple explanation for this distinctive feature of the study population. This was a very peculiar sample of patients. They were being studied at a time when they were still for the most part asymptomatic. Initially, these patients were too early in their illness to show any mortality. They were not really "terminal" at the time the study began. Mortality would only rise above zero when they became sicker. Because of the small number of subjects remaining at the time when the survival curve turned down, this inflection lacked statistical significance.

At one point, Pauling made the comment "...this study provides no information about the value of a continued intake of 10g per day of vitamin C in extending survival time, because none of the patients died while taking the vitamin.", (Pauling and Herman, 1989a). Figure 2 in Moertel et al (1985) does show the death rate in the ascorbate group increasing precipitously shortly after the median time for progression and discontinuation of treatment in the ascorbate group. However, this really doesn’t tell us anything about the relationship between individual patients discontinuing ascorbate and the time when they died. It is impossible to know from looking at the graph whether it was the same ascorbate patients who died and who had recently discontinued ascorbate.

Moertel (1986) having access to the data, was able to reply that the duration of vitamin C treatment was unrelated to any subsequent survival advantage. Moertel also notes, "Survival distributions measured from the last dose of vitamin C or the last dose of placebo completely overlap."

As long as we accept the statements of Dr. Moertel, the the rebound hypothesis is not feasible.

Leaving aside different effects on survival, the overall picture of ascorbate effects reported at the Vale of Leven is extremely different from that reported by the Mayo Clinic. This supports the idea that the situation in the Mayo Clinic studies differed in some fundamental way from that at the Vale of Leven. The patients at the Vale of Leven showed remarkable palliative effects in response to ascorbate, while these effects only appeared in a very minimal way at the Mayo Clinic. There were also reports of lethal reactions to ascorbate among four of the 50 sequential ascorbate treated patients described in Cameron and Campbell (1974). These patients showed fevers, necrotic hemorrhages of their tumors, and signs of intracranial hemorrhage. We are encouraged to accept the reality of these dramatic effects, because reports of this kind contradict the author’s hopeful expectation of benefits from ascorbate. Indeed, it seems difficult to believe that the authors could have invented anything so improbable. Nothing remotely like these effects was reported in the Mayo Clinic trials. This suggests that the Mayo Clinic studies failed to reproduce the situation that existed at the Vale of Leven.

One reason Cameron responded with incredulity to the Mayo Clinic’s data about urinary excretion of ascorbate, was that his own patients showed much lower levels of ascorbate in their urine, their plasma, and their leukocytes, even when receiving megadoses of vitamin C (Cameron and Pauling, 1993, pp. 122-6). As mentioned in the preceding chapter, there are numerous similar reports of this kind. These say that cancer patients often have very low ascorbate levels, that it may be difficult to raise these levels and difficult to produce an overflow into the urine. The differences in ascorbate level and the response to supplemental ascorbate, suggest some fundamental difference in the state of the patients at Vale of Leven and at the Mayo Clinic. This could be a reflection of the relatively minor ascorbate deficiencies in the Mayo Clinic patients and the major deficiencies at the Vale of Leven. This seems very natural when we consider that the subjects in the second Mayo Clinic study were not debilitated terminal patients, but were early in their illness at the time that they were being treated.

In a letter to the editors of the NEJM after the first Mayo Clinic study, Cameron (1980) had noted that no differences had been found at the Mayo Clinic in the survival times of ascorbate and placebo subjects, while both had survived longer than subjects who chose not to participate in the study. He suggested that this was due to the fact that the placebo subjects were ensuring their own supply of vitamin C by running down to the corner drugstore. At the time there were no published values for urinary excretion of ascorbate in the Mayo Clinic subjects.

In their reply, Creagan and Moertel, (1980) first pointed out that the group of patients who had declined to participate in the study had been in markedly poorer condition than those who did participate. Poorer survival in this group reflected poorer initial condition. They also described data from urine testing of a small random sample of nine patients from the first Mayo Clinic study. The 24 hr urine values for the four ascorbate subjects were said to be very high and those for five placebo subjects were said to be very low. Unfortunately, this statement was not accompanied by presentation of any actual data.

The second Mayo Clinic study also measured urinary excretion in a sample of 11 subjects. All of 5 ascorbate treated patients had 24 hr values equal to, or greater than 2 grams. Five of six placebo patients had levels equal to or less than .55g, while a sixth placebo patient had values intermediate between the treated subjects and the other placebo subjects. It was suggested that this individual’s values might be explained by the medication they were taking, which might interfere with the lab test.

There is a problem with these numbers. According to the literature (reviewed in chapter 1), the .55g/24 hours, is an absurdly high value for those not taking an ascorbate supplement. Something like 40 mg per 24 hours is much more likely for normal unsupplemented individuals, with lower values expected for cancer patients. Clearly these values (or at least those for the two high excreters in this group) were strangely high compared with those for the patients that Cameron had treated. Cameron and Pauling used this information to conclude that Moertel et al.’s placebo subjects must have been taking ascorbate on their own. Somehow, the Mayo Clinic workers never seemed to grasp that there was a problem with these values. Moertel (1986) stated that in the second Mayo Clinic study there was a clear difference in the urine levels of the ascorbate and placebo subjects, but again, failed to acknowledge that .55g/day for placebo subjects is anomalously high.

Hoffer(2001) has suggested that the Mayo Clinic studies would have lacked the statistical power to reliably indicate the beneficial ascorbate effects described by Cameron. This was because only a minority of Cameron’s patients showed dramatic improvements, and dramatically increased survival times. However, this does not seem an adequate reason to dismiss the Mayo Clinic findings.

The three studies carried out at the Mayo Clinic employed a total of over 300 subjects. Any one of the studies individually might lack sufficient power, but the three studies taken together would not. In the second Mayo Clinic study, the treatment with ascorbate seemed to accelerate progression, which of course is not compatible with the thesis of an improved survival effect, obscured by random error. In that study, the authors specifically tested the probability that ascorbate really exerted as much as a 25% improvement in survival, and were able to reject it at p=.017.

It seems to me that there was something peculiarly mechanical in the thinking about methodological issues in relation to the vitamin C trials I have described. The negative studies were methodologically sound. The positive studies were methodologically unsound. The negative studies produced valid data while the positive studies offered invalid data. There was however, never a realistic demonstration that the very strong positive findings of the Vale of Leven studies could have arisen as artifacts of these methodological inadequacies.

Suppose we knew none of the details about the conduct of the Mayo Clinic and Vale of Leven studies. Suppose we knew only that the Mayo Clinic studies were methodologically superior. This is largely the way in which the debate about the evidence was carried out. The Vale of Leven studies were not properly randomized, so it was possible in principle that the group of subjects receiving ascorbate was initially healthier than the group which did not receive ascorbate. In fact, according to the Mayo Clinic workers, this must have happened. How else could the positive results at the Vale of Leven be explained since they did not appear at the Mayo Clinic.

Because there were evidently no real ascorbate effects (after all, they hadn’t appeared at the Mayo Clinic), and because the ascorbate group must have initially been healthier, and destined for a longer period of survival than the controls, the groups should have had essentially similar courses to their illnesses. They should both have experienced similar, gradual declines from first presentation of symptoms to death. The ascorbate treated subject would just have lagged the subjects selected as controls because they were selected at an earlier stage of their illness, or because in some other way, they had better survival prospects. Somehow subjects with these favorable survival characteristics had been preferentially selected for ascorbate treatment.

The problem with this picture, is that the clinical reports (Cameron and Campbell, 1974) do not describe disease in the ascorbate treated patients as following the same course as in the controls. On the contrary, the course of their disease was sometimes marked by dramatic reversals and these reversals were time locked to the administration of ascorbate. Not only had Cameron selected for treatment subjects who had a superior probability of surviving, but they were selecting subjects who would have spontaneous remissions following ascorbate therapy.

If the Mayo Clinic’s picture of reality was correct, these improvements in patient’s conditions, time locked to treatment should not have occurred. Those who have never read the clinical report (or who have never thought about what they read), have missed a major reason not to accept the Mayo Clinic’s view.

As was described above, Moertel et al (1985) accused Cameron and Pauling of having "selected and reselected the patients they decided to evaluate". The implication is that significant differences between treated and control patients only emerged after Cameron and Pauling had struggled to come up with treated and control groups which would show the effects that they were looking for. This seems an entirely invalid description of subject selection.

In selecting subjects for both their 1976 and 1978 reports, Cameron and Pauling used the original 50 patients reported in the 1974 paper (Cameron and Campbell, 1974), as half of their sample of treated patients. They had felt compelled to do this because this group of 50 consecutive patients had been described in detail in the 1974 paper and they thought inclusion of these subjects would be expected. In all cases, the task of finding additional treated patients and controls, was handed off to persons who worked independently. This was done to avoid any bias in selection by Cameron or Pauling.

The comment by the Mayo Clinic workers seems to imply that Cameron and Pauling endlessly sifted through the data until groups could be found which differed in the expected direction. But there were only a total of two selections. The reselection for the second, (1978), report was inspired by comments that some of the treated patients described in the 1976 report, who had relatively rare cancers, were not matched with appropriate controls. Only 10 of the original hundred treated patients were replaced. Those involved in selecting the new treated subjects were blind to the survival dates of treated subjects. A new set of controls was selected which overlapped with the group of controls used in the 1976 report.

While the ratio of mean survival time for treated, relative to control patients, was higher in the 1978 report than in the earlier report, the obvious reason for this was not the reselection of subject. Rather it was because by this time 100% of the thousand control patients had died and 8% of the 100 ascorbate patients were still alive. With some survivors in the treated group and none in the control group, the ratio of mean survival times would increasingly favor the treated group simply because of the passage of time.

We have discussed above why the report of within subject changes in response to ascorbate in the Vale of Leven study makes the supposition of selection bias less relevant or entirely irrelevant. We have also noted that there is little prospect of rigorously proving the negative, that bias was not somehow present in the selection of treated and control subjects. However, it is still of interest to consider the plausibility of this occurring and of its accounting for the reported differences between groups.

What magnitude of effects might selection biases reasonably produce. When selecting subjects from a population of "untreatable" cancer patients, would a bias in subject selection plausibly have produced a 10% difference in survival rates between groups, a 25% difference, a 50% difference? But in fact, by the time that a year had elapsed since subjects were classified as untreatable, (this was the time at which ascorbate treatment was begun), 22% of the treated patients were still alive, and .4% of the historical controls. By the time of the last report about Cameron’s work (Cameron and Pauling, 1978) eight of their hundred treated subjects were still alive, and none of the thousand historical controls. This small group of long term survivors had lived an average of 7.7 times as long as their controls. This suggests that either ascorbate treatment was having some effect, or that Dr. Cameron and the other doctors at his hospital who administered ascorbate had a quite uncanny ability to pick out survivors.

The question arises as to whether there had been any "untreatable" patients as long lived as Cameron’s ascorbate treated survivors at the Vale of Leven, prior to the institution of ascorbate therapy. If not, this would seem to clearly indicate that selection bias could not have brought all of the long lived survivors into the ascorbate treated group. It appears that ordinarily there were no such long lived survivors in the pool of possible controls. We can’t definitively say such controls never existed because we don’t have access to all the patient records at Vale of Leven. However, the fact that at the end of the study period there were still 8 patients alive among the hundred treated patients and none among the thousand controls, strongly suggests that the treated patients were displaying survival times not otherwise present in this population of patients.

As far as I can determine, no positive evidence was ever brought forth to support the existence of bias in the process of finding controls. If such bias were present, it ought to be detectable in the numbers which describe the progression of treated and control patients through significant dates: that of first contact with the hospital about their cancer, that of evaluation as untreatable (when ascorbate administration was begun), and date of death.

Suppose the patients who were selected to receive ascorbate were on the average healthier than controls at the time when they were judged to be untreatable. If such a difference existed it would most obviously arise from ascorbate patients being classified as untreatable earlier in the course of their illness. For instance, it has been suggested that the doctors administering ascorbate, eager to put it to use, would have a motivation to go ahead and classify patients as untreatable earlier than they otherwise would have (Jaffe, 1982). This motivation might have been greatest with patients who seemed healthiest and seemed to offer the greatest hopes for survival. This selection bias would be reflected in a shorter time between initial contact with the hospital and judgement of untreatability (and beginning of ascorbate).

However, Cameron and Pauling found that in none of the nine cancer type groups did the treated group differ significantly from the controls in days between first cancer associated contact with the hospital, and date of "untreatability" (Cameron and Pauling, 1978). On average 388 days elapsed between the treated subject’s first contact with the hospital and their classification as untreatable, and 322 for the controls. This difference, 66 days seems rather substantial and favors the controls who would have been on average 66 days earlier in the course of their illness than the ascorbate treated subjects, at their assigned date of untreatability.

When this difference is examined within each of the nine cancer type subgroups, it was found that for four of the subgroups, the difference was in one direction, and for the other five, in the other direction. There is no obvious relationship between the direction of this difference and the size of the survival advantage of the ascorbate treated group.

The relative survival of current patients, treated with some new therapy, and that of patients treated in years past by other methods (historical controls), will not simply reflect the differing efficacies of current and past treatments. It will also reflect all the differences which have accumulated between two periods of time. Samples of patients from different periods might differ systematically in their experience of other treatments besides the two that are supposed to be compared, in the stage at which they are diagnosed, in their diet, in their ethnic composition, nutritional habits, socioeconomic status, etc, etc. This is why historical controls are regarded with such distrust.

Moertel et al speak of Cameron and associates comparing their patients with a historical control group. But in fact a considerable proportion of their control group had disease which was entirely concurrent with that of their matched treated patient. Cameron and Pauling (1978) made separate comparisons of their treated patients with contemporary controls (who had been hospitalized during the same time period), and with historical controls (who had been hospitalized before ascorbate therapy had been begun). The differences between the treated group and each group of controls was similar.

Here we will use the term "placebo effect" to refer only to effects on patients. The effects of hope and expectation on clinicians will be discussed in the section below.

While we cannot rigorously eliminate the possibility of placebo effects in producing improvement in relatively subjective symptoms, and in the patients’ general sense of well being, there are many indications that placebo effects were not significant at the Vale of Leven.

The palliative effects of ascorbate did not appear with the beginning of ascorbate administration, but only after a delay of some days. It appears that the time of their appearance is usually in the range of 5 to 10 days after beginning ascorbate. As Cameron pointed out, this delay is not at all what would be expected with a placebo effect.

Although we have only Cameron’s account to go on, his description of he situation at the Vale of Leven hardly suggests a large role for placebo effects. He states that the benefits of ascorbate were never described to patients or their families except in a very cautious way, as something "which might help a little", (Cameron and Campbell, 1974). Because Cameron seems to have always been an extremely scrupulous and cautious person, this account must be taken seriously. Cameron also says that intense skepticism about the possible benefits of ascorbate was the prevailing attitude among the staff until ascorbate had been used for some time.

The long duration of the palliative effects seen in some patients also does not suggest an explanation in terms of placebo effects. Nor does the fact that in many cases, the improvements in relatively subjective symptoms like mood and appetite were accompanied by more objective changes such as reductions of hematuria, ascites or malignant effusions. The existence of a plausible biochemical basis for some of the symptomatic responses to ascorbate (e.g. treatment of severe ascorbate deficiency, restoration of carnitine levels, altered prostaglandin metabolism), .makes it less necessary to invoke placebo effects in explaining the improvements seen.

Very similar kinds of symptomatic improvement were reported in the original studies of ascorbate’s palliative effects (in the 1940s and 1950s) and at the Vale of Leven. There is also the question whether placebo effects alone could be strong enough to account for the pain reductions reported in subject with bone metastases, some of whom were able to temporarily discontinue the use of opiates with ascorbate therapy.

In one of the sections above, we talked about the utterly different picture of ascorbate response which emerged at the Vale of Leven and at the Mayo Clinic (apart from the numbers that measure survival). In part these differences seem due to the careful clinical observations of the doctors at the Vale of Leven. The Mayo Clinic authors did not report the results of any comparable observations of their patients. Perhaps this is because doctors whose patients were in a double blind study would not be in a position to form opinions about how they were reacting to treatment. It seems plausible that doctors participating in such a double blind study would not be attentive to possible ascorbate effects in the same way as the doctors who knew what treatment their patients were receiving. In the second Mayo Clinic study, the fact that most of the patients were asymptomatic at the time the study began, would also minimize the appearance of symptomatic improvement with the administration of ascorbate.

While some of the information about the response to ascorbate at the Vale of Leven could properly be called objective (e.g. values from lab tests, tumor dimensions measured on radiographs), much of it clearly was not (e.g. statements about patients’ mood or appetite). The observations do not come from people who are blind to the treatment status of the patients. Must all the observations from the Vale of Leven therefore be dismissed as hopelessly subjective, anecdotal, and the not the product of a modern scientific methodology?

It is easy to imagine perception of the patient’s mood and general condition being distorted by the hopes, expectations, and wishes of the observing clinicians. On the other hand, there are many reported effects of ascorbate which simply don’t seem to be explainable in this way. The idea that Cameron and his coworkers could have simply imagined everything that they reported seems to have no psychological plausibility at all. Cameron and his coworkers were certainly capable of enthusiasm, but there is no reason to think that they were hallucinating.

A striking example is found in the case history of subject A (Cameron and Pauling, 1993, pp. 146-7). In this patient, large inoperable tumor masses had been found in the pelvis and abdomen when exploratory surgery was done. Later, and just before ascorbate was begun, the large tumor masses were still perceptible after fluid was drained from the patient's abdomen. With the administration of five grams a day of iv ascorbate, the patient soon showed a dramatic improvement. She was able to leave her bed, and the tumor masses shrank until they were no longer palpable. About a month later, the patient died. At autopsy it was found that death was due to intestinal obstruction, but that most of the tumors had disappeared.

So what happened here? How did this account come to be written in the absence of real anticancer effects exerted by ascorbate?

Did a placebo effect cause the patient to loose her tumors because she believed so intensely in the value of vitamin C? I don't think many oncologists would be happy with such an explanation even if the patient had believed fervently in vitamin C. But everything we know about patient attitudes at the Vale of Leven tells us that no such faith in the powers of ascorbate had developed among the patients, and especially not at the early stage when this patient was treated.

Perhaps the overheated imaginations of Cameron and his associates generated the entire story? They may have been capable of enthusiasm, there is no reason to think that they were insane. In fact, for the most part Cameron’s enthusiasm was a muted, cautious, slow developing thing. He was according to all the evidence a deeply cautious, a deeply conservative person. Even his years of struggle against orthodoxy couldn’t quite radicalize him. It seems to me unfortunate, that the combination of ascorbate with other agents, which might have dramatically increased its effectiveness was never appropriately pursued by Cameron and Pauling. And it seems to me that this had everything to do with caution, conservativism and lack of imagination.

As has been pointed out already, some reported ascorbate effects (rapid, lethal, development of high fever, tumor necrosis and hemorrhage) were certainly not part of any clinician’s rosy hopes for ascorbate. These reports are so surprising they could scarcely have been invented by Cameron and his coworkers.

It seems to me that the foes of ascorbate therapy, are begging some extremely interesting and important questions when they dismiss all of the clinical reports from the Vale of Leven as delusion and fantasy. These are supposed to be the inevitable result of the unscientific methods of Cameron and his associates. But what exactly is it that clinicians cannot be trusted to correctly perceive because of their beliefs and expectations? Would they simply imagine something like the case history of subject A?

It is interesting to consider the absence of clinical observations of ascorbate’s effects from the Mayo Clinic studies. The 1974 report from the Vale of Leven provides much richer information about the course of individual patients’ illness than do reports from the Mayo. One receives the impression that the clinicians at the Vale of Leven were deeply involved in observing the responses of their ascorbate treated patients, and perhaps at Cameron's smaller hospital, generally received more physician attention than at the Mayo Clinic. It seems plausible that the Mayo Clinic doctors, who were blind to treatment status, would be much less alert to changes in their patients following onset of treatment. The formal collection of data about patient symptoms took place only every two weeks in the first Mayo Clinic study, and much less frequently in the second. It seems plausible that the Mayo Clinic doctors might know vastly less about the effects of ascorbate on their patients than the doctors at the Vale of Leven.

Summary of Human Studies of Ascorbate and Cancer

In this and the preceding chapter we have reviewed almost all of the published data describing human trials of ascorbate and cancer. Below we recapitulate important points and present some conclusions. Some of the evidence referred to below, is described in chapters not yet published on this web site.

A detailed look at attempts to treat cancer with ascorbate, and at the entire range of evidence about ascorbate and cancer, provides strong reasons to believe in ascorbate’s anticancer effects.

Were the Vale of Leven trials methodologically sound? Was the study design such as to exclude the possibility of bias in the assignment of subjects to treated or control conditions? Clearly the answer to both of these questions is "No". So doesn’t the general failure of the anticipated ascorbate effects to appear in the methodologically sound Mayo Clinic studies prove that they didn’t really appear in the methodologically unsound Vale of Leven studies? The answer, in view of the entire range of relevant evidence, is again clearly "No".

An extensive background of evidence supports the general plausibility of ascorbate anticancer effects. Internal evidence from the Vale of Leven trials strongly contradicts the explanations suggested by the Mayo Clinic workers to discredit the Vale of Leven results. There is a range of differences in the particulars of the Vale of Leven work and the Mayo Clinic studies (e.g. nature of patient populations and form in which ascorbate was administered)  that might account for differnt outcomes. However, it does not seem that the use of intravenous ascorbate in the Vale of Leven studies is what distinguishes successful from unsuccessful use of ascorbate.

In a sense, it is absurd that we are still weighing the evidence from Cameron’s original studies, from the handful of other positive studies, and from the Mayo Clinic studies. The essence of science is supposed to be replication. Since there are unanswered questions about these studies, about what ascorbate can do for cancer patients, new studies need to be carried out. In some ways the decision to do such studies in terminal cancer patients should be exceptionally easy.

The agent being tested is familiar, safe, and inexpensive. As will be shown in chapters 7, the occurrence of procarcinogenic effects is likely to occur only under certain understood conditions which are avoidable. In view of the relative risks and benefits, terminal cancer patients surely have more to gain than to lose from a vitamin C study. There are positive studies which have shown why the goal is worth pursuing. There are negative studies which suggest what doesn’t work and what should be avoided in clinical trials. So, for instance, ascorbate should preferably be administered in the form and dosages used by Cameron and associates rather than those used at the Mayo Clinic.

Even if a full replication is not to be done, it would be of interest to measure whether the ascorbate/sorbitol solution used by Cameron produces significantly higher ascorbate levels in plasma or leukocytes than an equivalent amount of ascorbate taken in capsules. This should represent a very minor research effort and a starting point for further work.

Because of the recent positive study of ascorbate's anticancer effects in vitro by a pillar of the establishment (Chen et al, 2005) it now appears that there will be new studies with vitamin C in human patients. These studies will likely make use of very high iv doses, and ignore the kind of oral therapy used by Cameron.