J. Theor. Biol. 67: 625-635, 1977;
reprinted in Advances 1(3): 53-59, 1984.

A Theory of Diagnosis for
Orthomolecular Medicine


Molecular Disease Institute, Agoura, California, U.S.A.

(Received 27 October 1975, and in revised form 18 June 1976)


  It is assumed that most diseases arise from multiple causes, and that
diseases have the characteristics of polythetic classes. The signs and
symptoms of clinically-apparent disease are epiphenomena, or emergent
properties from the interaction among multiple biochemical etiologic
factors, intrinsic and acquired. Each individual carries a unique set of
intrinsic biochemical defects that are subsets of diseases to which he is
predisposed. He acquires additional defects throughout life. Such bio-
chemical defects can be detected by laboratory testing.

  Clinically-apparent diseases are sets consisting of multiple laboratory-
test anomalies associated with clinical signs and symptoms. Smaller sets
are formed by laboratory-test anomalies pertaining to the functional state
of major organ systems, without localizing signs and symptoms. The
latter sets are termed preclinical disease. Small sets of laboratory-test
anomalies, reflecting mainly intrinsic (genetic) defects, represent poten-
tial disease. Under appropriate conditions, elements can be added to or
subtracted from the sets, so that diseases may evolve to a more advan-ced stage or regress under therapy. Ideally, sets of biochemical anomalies
should be identified at an early stage, before expansion of the sets even-
tuates in clinically-apparent disease.

1. Introduction

  Orthomolecular medicine is defined as the provision of the optimum
molecular constitution, especially the optimum concentration of substances that are normally present in the body, for the purposes of treating disease and preserving health (Pauling, 1968, 1974). Pauling's innovative concept has had scant influence on the practice of clinical medicine. Major impediments to acceptance of the orthomolecular concept appear to be difficulties in identifying molecular lesions and uncertainties over the meaning and measurement of optimal. Consequently, it is difficult to select appropriate therapies and monitor therapeutic responses with confidence. In general, the few clinicians who practice orthomolecular therapy tend to assume that the existence of a given clinical syndrome implies a particular biochemical pathosis, in much the same manner that wet streets imply rain. Thus orthomolecular medical practice is hindered not only by diagnostic imprecision , but also by a post hoc fallacy that rests on traditional (and, it will be shown, antiquated) diagnostic methodology.

  The human organism can be regarded as a set of subsystems, of which integrated functional measurements can serve to predict the status of the organism as a whole (Patton, Huemer, Hussman & Caines, 1963). Along functional lines, subsystems may be identified as assimilatory, excretory, regulatory and circulatory; or, according to a more traditional schema, digestive, renal, neuroendocrine, cardiovascular, etc. Such subsystems are further divisible into groups of cells performing specialized functions characteristic of one (or more) subsystems. In each group of cells is expressed that portion of the organism's genome that is relevant to the specialized functions. Differing groups of cells possess similarities as well as differences among their complements of active genes, and thus possess similarities as well as differences in their metabolic patterns. This fairly conventional way of regarding the living organism can serve as the departure point for novel insights into the nature of disease and the method of its diagnosis.

2. Nature of Disease

  A disease is an abnormal functional state characterized by a set of symptoms and physical findings. Diseases as seen in individual patients often depart from textbook descriptions of disease in that some of the characteristics may be lacking, or the characteristics may be associated with those of other disease states. Diseases possess the characteristics of polythetic classes*. According to Sokal & Sneath (1963), a polythetic class fulfills the following criteria: (a) each member of the class possesses a large (but unspecified) number of the properties that characterize the class as a whole; (b) each property of the class is possessed by large numbers of the class members. In addition, a class is deemed fully polythetic if no property of the class is possessed by every member of the class.
*The term "fuzzy sets" has a similar meaning.

  Disease classes are polythetic with respect to etiology as well as symptomatology. It is now generally understood that many--perhaps most--diseases result from multiple causes. A class such as tuberculosis is not fully polythetic because all cases possess a common etiologic factor (the tubercle bacillus) in addition to various other etioiogic factors. However, some of the chronic degenerative diseases may be fully polythetic with respect to etiology, which may explain why their control is elusive; there may exist no single point at which the process can be controlled. To restate the matter, the organism can become diseased for a variety of reasons, but there may be only a limited spectrum of possible dysfunctional responses to an immense number of possible combinations of causative factors.

 Since the living organism is a biochemical mechanism, the causes of dys-
function must at their most fundamental level be biochemical in nature. In
the final analysis, except for traumatic conditions, all diseases become
molecular, whether the defects are congenital or acquired (Huemer, 1972).
Some biochemical lesions are acquired during life when environmental
stresses (e.g. toxins, malnutrition, infection) exceed the reparative capacities of the living system. Others are intrinsic, or genetic. It will be recalled that groups of cells possess differences from and similarities to each other with respect to their biochemical patterns and active-gene complements. A set of genetic defects might affect one specialized group of cells more than others, but would probably affect many cell groups to some degree. The polythetic arrangement of gene-expression sets is echoed in the polythetic nature of disease itself.

3. Genetic Substrate of Disease

 From the outcome of consanguineous matings it appears that the average
human being is heterozygous for one or two genes that would cause serious
disease in the homozygote (Carter, 1967). Muller (1950) estimated that the
average human is heterozygous for a minimum of eight, and possibly scores
of genes, each of which produces a slight detrimental effect. According to
Muller, the combination of such genes, acting together, gives the individual
his own characteristic pattern of weakness. More recently, Hubby & Lewontin
(1966) demonstrated a previously unsuspected degree of polymorphism
among the genes of Drosophila pseudoobscura in wild populations; at least
39 % of the loci in the genome are polymorphic over the whole species, and
an estimated 8 to 15% of an individual fly's genome is in the heterozygous
state. If the finding may be generalized to humans--and there is no compelling
reason to suppose otherwise--then Muller's estimate for prevalence of sub-
vital genes seems conservatively low. Furthermore, any individual will be
homozygous for numerous isoalleles that may confer reduced fitness only
in the homozygous state.

 Along related lines, the enzyme studies of Harris (1966) led him to con-
clude that enzyme polymorphism may be a fairly common phenomenon
among the very large numbers of enzymes that exist in humans. The degree
to which enzyme polymorphism affects health is not presently known. It
appears quite probable, at any rate, that the average human individual
carries thousands of isoalleles, and it is reasonable to suppose that many of
the isoalleles are subvital to a degree. Thus the individual will carry through-
out life a characteristic pattern of molecular weaknesses.

 In principle it should be demonstrable that an individual carries a unique
pattern of biochemical anomalies that persists, with variable degrees of
expression, for many years. The available data on normal characters (Cotlove,
Harris & Williams, 1970) indicate that an individual's biochemical measure-
ments generally remain quite stable over a span of many months, fluctuating
narrowly about homeostatic set-points. A similar study seems not to have
been conducted on abnormal characters, but this author has observed a
27-year-old Caucasian male who had fasting hyperg]ycemia, a-lipopro-
teinemia, relative lymphocytosis, high salivary uric acid, high salivary
cholesterol, and high salivary chloride; salivary ascorbic acid was abnormally
low. The patient had previously been examined at the ages of 12, 14 and 16
years. He had a relative lymphocytosis on all three occasions, low salivary
ascorbic acid on two occasions (borderline-low on the third), and high
salivary cholesterol on two occasions; high salivary chloride and borderline-
high fasting glucose had been found on the single prior occasion when those
tests had been performed. Relevant to a possible genetic origin of the
anomalies, it should be noted that all of the findings except fasting hyper-
glycemia were present in one or both parents, and one parent has an aberrant
glucose-tolerance curve.

4. Classification of Disease

 The objection might be raised, in relation to the referenced clinical-
laboratory studies, that the tests provide only indirect measurements of
genetic qualities, and probably reflect polygenic traits; hence they do not
necessarily indicate potential molecular causes of any disease but instead may
be effects of underlying genetic pathology. Furthermore, since most clinical-
laboratory measurements are made under homeostatic conditions, they will
fail to detect numerous weaknesses (reduced capacity for enzyme synthesis,
for example) for which the organism is able to compensate in the unstressed
state. The objections are valid in so far as they apply to any attempted use
of such data to classify disease along strictly etiologic lines. At present, it is
not feasible to measure large numbers of direct gene-products in the intact
organism; hence a strictly molecular-etiologic classification is not presently

 However, it is not necessary to know cause-and-effect in order to classify
diseases by biochemical and other laboratory-test criteria. It is not necessary,
for example, to know whether hyperuricemia originates in a hereditary
enzymic defect or as the result of ingestion of drugs or excessive beef, in
order to state that hyperuricemia belongs to the class, gouty arthritis. It is
not necessary to explain the absence of hyperuricemia in some cases of gout
or its presence in other diseases, in order to place hyperuricemia in the gouty-
arthritis class. For purposes of disease classification, laboratory-test anomalies
possess the same significance as clinical signs and symptoms, with this
important difference: laboratory-test anomalies are subsets of signs and
symptoms, and are less far removed than clinical signs from the ultimate
molecular causes.

 Sokal & Sneath, in suggesting the application of polythetic concepts to
the construction of disease taxa, have commented that etiology may be
unsuitable as a general principle for defining taxa (Sokal & Sneath, 1963,
p. 283). However, a truly etiologic classification is theoretically possible and
will surely be achieved in time; meanwhile, a polythetic biochemical-phenetic
classification should serve as a useful approximation to that ideal.

 A few attempts have been made at mathematical diagnostic interpretation
from multiphasic biochemical test data. Thus, Pomeroy (1975) reports
statistically significant separation of hamsters into original vendors' groups
by means of computerized linear discriminant analysis of plasma amino acid
profiles. Using computerized pattern-recognition calculations, Robinson &
Pauling (1974) have demonstrated correlation of chromatogram patterns
with sex, fasting, birth-control medication and neuromuscular disorders. A
criticism of such work is that the experiments are usually not based on
known biochemical mechanisms related to particular diseases, so that it is
difficult to reason inductively from the data. By ignoring mechanisms and
limiting the methodology to traditional, clinically-established diagnostic
categories, one invites difficulties in dealing with undifferentiated early
disease, transitional states, semantically imprecise categories, and multiple
simultaneous diseases.

 Implicit in the preceding discussion is the assumption that normality can
be defined; otherwise, anomalous has no meaning. Many physicians believe
that each individual possesses his own "normal" set of biochemical test
values, by which it is meant that the individual's health is good when various
tests give certain numerical results, whether or not those results fall within
the statistically normal range. I reject this arbitrarily relativistic view.
Normality, in the context of this discussion, means normally distributed
about a mean; in clinical laboratory medicine, the normal range of a para-
meter is plus or minus two or three standard deviations from the mean of a
healthy population (Amador, 1975; Copeland, 1972; Files, Van Peenen &
Lindberg, 1968). Strictly speaking, the distribution of test values in healthy
persons is not Gaussian, as Elveback, Cuillier & Keating (1970) point out,
and the false assumption of a Gaussian distribution can lead to mis-
classification of some test results as "normal". Regardless of the chosen
statistical criteria, it seems clear that most or all of the "normal range"
corresponds to that condition which is termed by geneticists the "wild type",
symbolized by +. Deviations from the wild type are not manifestations of
individual good health, but of single or multiple genetic defects (including
associated compensatory mechanisms). Cheraskin & Ringsdorf (1973)
analyzed the mean and variance of fasting blood-glucose values for 100
dental patients. By selecting the patients within this group according to
progressively more strict criteria for oral health, they achieved, by stages, a
slight reduction in the mean and a tenfold restriction in the normal range.
It is reasonable to suppose that laboratory-test values should be standardized
on populations meeting only the most stringent criteria for health and vigor
(cf. Files et al., 1968; Cheraskin & Ringsdorf, 1973).

5. Evolution of Disease

  Clinically-apparent disease usually has evolved through stages of potential
disease and preclinical disease, the latter being a dysfunctional state of one
or more major subsystems that does not produce a characteristic set of
symptoms and signs. In the beginning exist isolated biochemical anomalies
(diatheses), representing the individual's unique genetic pattern. Each
diathesis may be regarded as a potential disease. As further anomalies
accumulate (resulting from the interaction of environmental stresses and
constitutional weaknesses), the stage of preclinical disease is attained. This
tendency to acquire biochemical anomalies with advancing age shows up
as an age-related broadening of distribution curves for laboratory test
values, as observed by Files et al. (1968). Possible mechanisms include exo-
genous agents (toxins, viruses), age-related enzyme polymorphism (Gershon
& Gershon, 1973), and failure of homeostatic feedback loops (as in the post-
menopausal deregulation of FSH secretion). Finally, the accumulation of
still more anomalies results in frank symptoms and physical findings. To
put the matter in symbolic terms, let X be a polythetic class (disease)
characterized by the anomalies m, n, o, p and q. At time T0 the individual
possesses m and n, plus the unrelated anomalies t and w. As mntw do not
form the major part of a disease set, the individual is apparently healthy.
At T1, q is added to the individual's set. The individual at this point may feel
vaguely ill, as more of the set mnopq has been completed, and q may perhaps
be part of another set that has also been brought nearer to completion.
Finally, at Tx, o and/or p occurs, and the individual experiences a case of X.

visual analogy

Website Design, Hosting and Maintenance by User-Friendly Computing
505 River Street, Santa Cruz, CA 95060  • Tel:  (831) 423-9653
Copyright (C) 2008.  All Rights Reserved.