Analysis Of A Survey of Patients
On Immuno-Augmentative Therapy
Robert G. Houston
Few established scientists have seriously examined the experience
of patients on alternative cancer therapies. An outstanding exception
is Barrie Cassileth, Ph.D and her team of researchers at the University
of Pennsylvania Cancer Center. Their 1984 survey of patients and
practitioners using unorthodox cancer therapies (1) was a milestone
in the objective analysis of the subject, and has been acclaimed
by all sides as the finest study in the field.
It is thus significant that a new study from the same research
group reports surprisingly positive results in patients on an alternative
immune therapy for cancer. In September, 1987, pre-publication copies
of the manuscript were circulated for review. Entitled, AReport
of a Survey of Patients Receiving Immuno-augmentative Therapy,@
the study (2) concerns 79 cancer patients, most of them with metastatic
or inoperable disease, who were treated at the IAT Centre at Freeport
in the Bahamas, where the therapy of Lawrence Burton, Ph.D. is clinically
applied.
The survey (2) found that prolonged survival occurred in these
patients, averaging 5 years from initial diagnosis, with 63% still
alive at the time of analysis. Their expected survival time based
on tumor site and stage at diagnosis would have been less than 3
years, according to the researchers. After the first course of treatment,
37% became more ambulatory and 29% had improved appetite. Adverse
side effects were reported by only a ???? small percentage of the
patients. Pointing to possible methodological problems in the study,
the researchers called for careful clinical trials of the Burton
regime.
Because of the refusal of the FDA to permit clinical use or testing
of the treatment in the U.S., Dr. Burton was forced to move his
clinic from New York to the Bahamas in 1977. Immuno-Augmentative
Therapy (IAT) is regarded as an AUnproven Method@ by the American
Cancer Society (3), despite its reported production of rapid tumor
regressions in animals (4-6), which was publicly demonstrated by
Burton at the American Cancer Society=s Science Writers= Seminar
in 1966 (7). Officials of orthodox medicine have deprecated the
treatment as lacking published studies in peer-reviewed medical
journals. When Burton attempted to publish in 1963, 1967 and 1972,
however, the papers were rejected, discouraging further attempts.
The National Cancer Institute and the FDA, pursuing a standard policy
of opposition toward alternative cancer therapies, sought for years
to persuade the Bahamian authorities to close the clinic and succeeded
in doing so from July 1985 to March 1990 with allegations of serum
contamination (8). When these were later shown to be false and misleading
(9), the clinic was allowed to reopen. As a probable consequence
of decomposition of the materials due to NCI=s apparent failure
to keep the serum samples frozen, the NCI researchers (8) reported
they could find no active components.
General Concerns
With this background, the Cassileth team appear to be genuinely
embarrassed by the positive results of their survey. Most of their
paper is spent trying to discredit their own findings. Much is made
of suggested factors that might have biased the sample positively,
including age, race, ambulatory status, and socioeconomic class,
while important negative biasing factors, such as the failure of
conventional treatment on most of these patients and the 8 month
closure of the clinic, are ignored. It soon becomes clear that the
claimed positive biases, while mostly minor or irrelevant, are being
used - perhaps unintentionally - to discount not only the survey
results but also by extension any benefits that the clinic may achieve.
In this process, essential data that the authors indicate were
obtained (2:5), including subsequent medical status of the patients,
effects on tumors, extent of confirmation by conventional doctors,
and comparative results in matched controls, are not provided. Moreover,
the draft copy was flawed by miscalculations of means and medians,
and even most percentages - always in a direction that would diminish
the results. As a consequence, misinterpretations contrary to the
data are presented, which have the effect of playing down the strong
indications of therapeutic response.
It is proper and commendable for researchers to note the limits
and deficiencies of their study and to examine alternate explanations
for their results. Selection bias is a genuine problem in most medical
research, and the authors conscientiously point out the possible
ways in which it may have influenced their survey. While some of
their points are quite reasonable, others may be questioned.
Why No Controls?
Most of the problems of the study could have been obviated by appropriate
matching. Absence of a matched control group allows the authors
to dismiss the IAT results with fanciful hypotheses unsupported
by valid comparisons. They note (2:2) their original intent to compare
the IAT patients with comparable patients receiving conventional
treatment at the University of Pennsylvania Cancer Center. In a
letter of May 21, 1987 to the IAT Patients= Association, Dr. Cassileth
acknowledged, AWe are now working to find matched controls for each
patient (matched on specific disease and stage at diagnosis, sex,
age, group, race).@ What happened to this data? Was it simply too
difficult to find adequate matches, or was the data withheld because
of a dramatic difference between the groups?
The main reason provided for the Lack of comparative data is that
only 29 patients...met eligibility requirements (available biopsy
reports and metastatic disease at diagnosis)@ (2:2). Such a number
actually constitutes a sizable sample; Phase II trials in cancer
often involve only about 20 patients (10). Furthermore, if the cases
were to be matched, why must they be restricted to metastatic disease?
Another rationale, that there was unintentional selection bias,@
is actually a reason why matching should be employed, not avoided.
Its primary function is to compensate for selection bias in the
experimental group.
Finally, the authors assert that their main goal was only that
of encouraging patients to seek hepatitis and HIV testing.@ In
this goodwill effort, they assembled patients known to be alive,@
thereby producing an unintentional bias toward patients with longer
survival.@ (2:11). This undoubtedly is a valid point, and by far
the most important positive bias in the survey. It is counterbalanced,
however, by the fact that patients still alive at analysis, who
constituted a majority of the sample, were artificially considered
deceased when calculating the survival time. Thus, a living patient
who had been diagnosed one year before would be counted as if he
succumbed at one year, though he may in fact survive many years
more. It should also be noted that the survey began with 32% of
the group deceased, most of them from an earlier survey (1) conducted
in 1982-83 from which half of the cases came. Many of these deaths
may have occurred during the 8 month period when the clinic was
closed and the therapy withdrawn, producing another negative bias
in the study.
Mistaken Percentages
Throughout the draft report, important percentages were diminished
by including unknown cases in the denominators. Inclusion of unknowns@
has the effect of an assumption that they all lack the feature being
measured. Unless there is a sound explanation for believing so,
however, it is proper to assume that unknown cases have the same
distribution of features as the known cases. This is accomplished
equally well by excluding them from the calculation. The NCI follows
this conservative practice and excludes from its survival figures
unknown cases lost to follow-up (11). If cases of unknown status
are to be included in the figures, 3000 others could also be added
who were treated with IAT over the years.
Such errors are ubiquitous. In Table 3 for example, among the 70
patients whose results were known, 26 patients or 37% (not 33%),
had improved to ambulatory status. Among the 51 patients whose medical
status was known, 38 patients or 75% (not :48%@) had progressive
disease before starting IAT. An appropriate revision would eliminate
the percentages for rows marked unknown@ in the tables, and recalculate
all other percentages based on denominators that exclude the unknown
cases. If desired, separate tables or columns can be provided to
show the proportion of unknown cases without distorting the presentation.
Incidence of Infection
When it comes to figures that could put the treatment in a bad light,
the researchers are suddenly careful to exclude the unknown cases
and thus maximize the percentages. We are informed that Among all
patients tested, four of 23 patients, or third of the sample was
????l however, how can the question whether the results are representative:
they may have been tested because they were in a special risk category,
based on lifestyle or prior transfusion. Unfortunately, no other
information is given about the cases, or the type of tests, or whether
the results were confirmed, or even whether the tests the patients
mentioned had occurred after the start of IAT or before. The lack
of details is surprising inasmuch as information on safety is stated
to be the main purpose of the survey, contrary to its representation
to the cooperating patients.
Since 2 million adult Americans are estimated to be positive for
HIV antibody, the one positive case in this uncontrolled sample
means nothing regarding origin. In some U.S. cities, over 60% of
male homosexuals are positive for HIV antibody. Furthermore, positive
results in the standard ELISA test are false in 63-89% of the cases,
according to the AMA, and require confirmation by the more specific
Western blot test (12). When the samples of the Burton serum alleged
to be positive for HIV were tested by the Western blot, they were
all found to be negative (9), information that the NCI did not mention
when seeking to close the clinic (8).
According to the Centers for Disease Control (13), The estimated
lifetime risk of hepatitis B virus infection in the U.S. varies
from almost 100% for the highest-risk groups to approximately 5%
for the population as a whole.@ Among the groups with high rates
are patients who have had blood transfusions (14), hospitalization
and surgery (14), or cancer (15). According to the American Public
Health Association (16), Overall, 5% of the adult USA population
has anti-HBs@ (antibody to hepatitis B surface antigen). Cassileth
et al do not reveal whether the 4 positive patients were only antibody
positive, indicating past infection or vaccination and present immunity,
or demonstrated the rarer viral surface antigen indicative of current
infection. They also fail to mention whether any patient actually
developed clinical hepatitis.
They note that the patients in their sample who were tested subsequent
to the clinic=s reopening were all negative for hepatitis. It should
also be noted that sophisticated equipment and procedures have been
established by the clinic to ensure the safety of the serum. The
procedures are designed to eliminate viruses but not antibodies,
as these are non-infectious and are regarded as beneficial for immunity.
Gamma globulin, which contains both hepatitis B antibody (13) and
HIV antibody (17), is often given in conventional medical practice
to boost immunity and resistance to hepatitis (13). Those receiving
such materials may show up positive on routine antibody tests (13),
but the results would have no significance regarding actual infection.
Survival Time
In the draft copy of the Cassileth paper, the 6-month follow-up
column of Table 3 was extremely misleading, for 25 patients who
were deceased long before the survey even began were included, with
the implication that 37%@ died in 6 months. In fact, as noted elsewhere
in the paper (2:5), only 4 patients (7%) of the 54 who were followed
had died in the 6 month interim.
The 6 month mortality rate of 7% is remarkably low for patients
with advanced cancer. It is particularly significant as it was obtained
prospectively, with a follow-up 6 months after the initial interview.
Extrapolated to the future, it would yield a one-year survival rate
of about 86%, and a 5-year survival rate of about 46% from the initiation
of the Burton treatment: (50/54)~10 * 100 = 46%. Relative survival
rates, which adjust for other causes of death, are 20% higher, so
this would be equivalent to a 5-year relative survival rate of about
56%. Given the limited size of the sample and the brief follow-up,
this must be regarded as only a crude estimate, but is consistent
with the stated survival averages. A one year follow-up would have
been more valuable, but apparently was not conducted.
Data is given in Table 4 of the paper for mean and median survival
time from the initial diagnosis. Nowhere is it acknowledged that
these are minimum measurements of an increasing value. Since 63%
of the patients studied were still alive at analysis, the actual
average survival time in the group can be expected to be much longer
than the stated 5 years (mean = 63 months, SD = 38.7). A median
survival time cannot properly be calculated until half the patients
have expired; a valid median might have been calculated for the
34 patients from the earlier study, but the authors do not provide
the data. Also useful would have been separate figures for the survival
of the metastatic patients.
Because the standard deviation is given, the significance can be
calculated. At analysis, the mean survival of the 79 IAT patients,
most of who had advanced cancer, was already 75% longer than the
36 months maximum expected survival time (2:7). The difference is
highly significant statistically (t = 6.2, P < 0.00000001), with
odds against chance of 100 million to one.
False-Positive Biases
AMBULATORY STATUS. The authors hypothesize that these
remarkable results are due to a selection bias in that patients
who travel to the Bahamas would be more ambulatory. Wheelchairs,
crutches and stretchers are available for travel, however, and any
such bias would also apply to regional cancer centers in the U.S.
In Cassileth=s data, 17% the IAT patients (12 of the 70 with known
status) were non-ambulatory when presented for treatment. This in
fact is a higher proportion than usual. In a study (18) of the performance
status of 612 advanced cancer patients at M.D. Anderson Hospital,
10% were non-ambulatory (performance grade of 3 or 4 on the Zubrod
scale). In a Mayo clinic study (19) of 179 advanced cancer patients,
all the patients were ambulatory, yet median survival was only 5
months. Dr. Cassileth=s theory ignores such relevant comparisons
and is essentially nullified by her own data.
AGE. As a second excuse for dismissal, the authors
state that the age of the IAT patients is younger than average for
cancer patients. From the data in Table 2 concerning age at diagnosis,
however, it is evident that the authors have miscalculated the ages.
The actual median of this data is 61 years, not the 54-years@ they
give. (Since the 50% point of the sample falls in the 60-69 age
interval, it is not possible for the median to be less than 60).
According to the NCI (20), the average age of whites at diagnosis
of cancer is also 61. Furthermore, the patients went to the IAT
clinic an average of 17 months after diagnosis (2:7) and thus would
have been somewhat older on average than cancer patients starting
treatment in the U.S.
The purported mean age of 51 years@ is only possible through the
fallacious use of the low point, instead of the midpoint, to represent
each age interval. It would be a remarkable coincidence indeed if
all 28 patients aged 40-59 happened to be age 40, as such a practice
assumes. Since only 2 of the patients were under 21 (2:4) it is
virtually impossible for insufficient information is provided for
a valid estimate of the mean, it must be many years - perhaps a
decade -higher than stated, given the age distribution presented
(Table 2).
RACE. A third excuse is that the patients are all
white. For decades, however, the only national statistics on cancer
survival were limited to whites (21), and today the survival figures
available from NCI are usually given separately for blacks and whites,
rather than integrated (11). A racial excuse is thus irrelevant.
SOCIOECONOMIC STATUS: If patients with higher socioeconomic
backgrounds survive longer, perhaps it=s because they have the knowledge
and means to seek unconventional therapies such as IAT. The data
in Table 2 of the paper shows a normal proportion of homemakers,
but indicates that among 52 employed persons 17% were professionals
and 15% were blue collar workers. The 1985 figures for the U.S.
are 13% professionals and 31% blue collar (22). Cassileth et al
may well be correct in concluding that the patient sample reflects
a higher socioeconomic status. They overstate, however, the survival
implications.
The main proponent of the socioeconomic hypothesis in regard to
cancer survival is Dr. John Berg of the University of Colorado,
who points out that the effect is seen mainly in the very poor (23).
His own study concerning 5 year cancer survival rates in metropolitan
counties of Colorado found that with standardization for sites,
Hispanics as a group, though mostly poor, had slightly better overall
survival than high-income whites who were non-Hispanic. Furthermore,
between high income and low income whites (Anglos) the difference
in 5 year survival rates was only 5 percentage points (23). The
difference may or may not be significant, but even if real would
be rather minor.
The Forgotten Bias of Advanced Disease
The primary objection to the Cassileth report is that while it
reaches for spurious positive factors to discount the prolonged
survival results, it ignores the profoundly negative biases in the
sample. Central among these is the fact that most of the patients
came to IAT with advanced cancer that was refractory to conventional
treatment. They thus constitute a negatively skewed sample composed
mainly of the failures of orthodox medicine, in effect its most
difficult cases. This is corroborated by the fact that 75% of the
patients (38 of the 51 with known status) had progressive disease
prior to starting IAT (2:16), even though 86% of the patients had
completed conventional treatment as prescribed (2:6). Moreover,
58% of the patients (44 out of 76 known) had metastatic or inoperable
disease at diagnosis, and thus a poor prognosis. Because they were
17 months past diagnosis on average at arrival, time would have
almost run out for many and the disease of others would have reached
a more advanced stage by that point.
Median values for overall cancer survival are no longer readily
available from NCI. In 1972, however, the median survival time for
white cancer patients in general was given as 1.6 years for observed
survival, and 2 years for relative survival (21). By comparison
with this or with the 36 month maximum expected survival suggested
by Cassileth et al (2:7), the prolonged survival in the Burton patient
group was far greater than expected and strongly indicates therapeutic
efficacy, which the authors invalidly discount. Their paper needs
major revision.
Conclusion
Despite several reservations, it is clear that Dr.
Cassileth and her associates have produced an important, responsible,
and intelligent examination of the experience of those who have
ventured beyond the restrictions of cancer orthodoxy to obtain an
alternative therapy. In dealing with this controversial topic the
researchers have taken care to fully qualify their findings in a
conservative presentation. Their attitude and conclusions, however,
properly encourage clinical research rather than further suppression
of the Burton approach. It is hoped that their valuable study will
be properly corrected and published.
REFERENCES
1. Cassileth, B. R., Lusk, E. J., Strouse, T. B., and Bodenheimer,
B. J. Contemporary unorthodox treatments in cancer medicine. Ann.
Int. Med. 101:105-112, 1984.
2. Cassileth, B. R., Trock, B. J., Lusk, E. J., Blake, A., Walsh,
W. P., and Arnholz, D. Report of a survey of patients receiving
immunoaugmentative therapy. :Philadelphia: Univ. Of Pennsylvania
Cancer Center, 1987.
3. American Cancer Society. Unproven Methods of Cancer Management:
Immuno - Augmentative Therapy. Cs 34:232-237, 1984.
4. Burton, L., Friedman, F., Kassel, K., Methods for determination
and alteration of titers of a complex of factors present in blood
of neoplastic mica. Trans. N.Y. Acad. Sci. 25:33-38, 1962.
5. Kassel, R., Burton, L., Friedman, F., Harris, J. J., Synergistic
action of two refined leukemic tissue extracts in oncolysis of spontaneous
tumors. Trans. N.Y. Acad. Sci. 25:39-44, 1962.
6. Friedman, F., Burton, .L., Rottino, A. Necrosis, liquefaction
and absorption of CCH mammary tumors resulting from injection of
extracts from tumor tissue. Proc. Am. Assoc. Cancer Res. 6:20, 1965.
7. Yasgur, S. S., Can cancer be destroyed by the body=s own agents?
Modern Medicine, Jan. 1, 1975, pp. 40-45.
8. Curt, G. A., Katterhagen., G., and Mahaney, Jr., F. Immunoaugmentative
therapy: a Primer on the Perils of Unproved Treatments. JAMA 255:505-507,
1986.
9. I.A.T. Patients= Association. Rebuttal to JAMA. Health Consciousness
7 (4):6-9, Aug. 1986.
10. Sylvester, R. Planning cancer clinical trials. In: Cancer Clinical
Trials. Eds.: M.S. Buyse, M. J. Staquet, R. J. Sylvester. Oxford
Univ. Press. 1984, pp. 47-63.
11. ????, L. G., Pollack, E. S., and Young, J. L. Cancer patient
survival: SEER Program, 1973-1979. JNCI 70:693-707, 1983.
12. American Medical Association, Council of Scientific Affairs.
Status report on the Acquired Immunodeficiency Syndrome. JAMA 254:1342-1345,
1985.
13. Centers for Disease Control. Recommendations for protection
against viral hepatitis. Ann. Int. Med. 103:391-402, 1985.
14. Aach, R. D., and Kahn, B. A., post-transfusion hepatitis: current
perspectives. Ann. Int. Med. 92:539-546, 1980.
15. Rosenthal, S., Carignan, J. R., and Smith, R. D. Medical Care
of the Cancer Patient. Phil: W. B. Saunders Co., 1987, p. 410.
16. Benenson, A. S., Control of Communicable Diseases in Man. Washington:
American Public Health Assoc., 1985, p. 172.
17. Steele, D. R. Letter to editor. JAMA, Feb. 7, 1986.
18. Swenberton, K. D., Legha, S. S., Smith, T., et al. Prognostic
factors in metastatic breast cancer treated with combination chemotherapy.
Cancer Res. 39:752-758, 1979.
19. Moertel, C. G., Fleming, T. R., Rubin, .J., et al. A clinical
trial of amygdaline (Laetrile) in the treatment of human cancer.
New Eng. J. Med. 306:201-206, 1982.
20. Axtell, L. M., Asire, A. J., Myers, M. H. Cancer Patient Survival:
Report No. 5. National Cancer Inst. 1976, pp. 3-4.
21. Axtell, L. M., Cutler, S. J., Myers, M. H. End results in Cancer:
Report No. 4. National Cancer Inst. 1972, pp. 1-3.
22. Hoffman, M. S. (Ed.) The World Almanac 1987. New York: World
Almanac. 1987, pp. 128-129.
23. Berg, J. W. A comparison of survival rates between indigents
and non-indigent populations. In: Cancer Treatment and Research
in Humanistic Perspective. Eds.: S. C. Gross and S. Garb. New York:
Springer Publ. Co., 1985, pp. 104-117.
