What is precision treatment and how did it change? Treatment follows an evolutionary course.
Syphilis was treated by mercury, then Salvarsan then Penicillin G.
Dropsy by leeches, then Digitalis, now treatment is aimed at underlying causes: IHD, Valvular disease, cardiomyopathy, hypertension, diabetes mellitus.
Diabetes mellitus by a low-starch diet and vigorous exercise, or an herbaceous blend of lupine (a legume with edible peas and gaudy flower spikes), fenugreek (a small herb with pungent yellow seeds) and zedoaria (a wetland crop whose roots taste like ginger with a bitter aftertaste, then insulin. Progress to GLP-1 for type 2 diabetes
All these are increases in diagnosis and treatment: This is precision medicine during the last 2.5 millennia. Thus the approach is same wine, new bottle.
Today the leading contenders for inclusion in precision medicine are suggested by biogenetics, microbiomics, epitope alterations, lifestyle, co-existing disease, ethnicity, sex, age, the factors that most physicians practicing medicine would consider in order to diagnose and treat disease. It is a holoscopic rather than meroscopic approach, as in precision medicine the use of artificial intelligence, relying on AI and a mechanistic appropriation of data points. But diseases evolve so population-based data require real-time processing as evolutionary changes occur.
Is precision medicine’s mechanistic approach sufficiently different to change medicine ? Has not the increase of specific treatment always been the aim? Is this not historically the case?
Consider complex systems; ‘the butterfly effect’ Would the microbiome or the co-existence of depression, hypertension or diabetes influence this complex system? Would precision medicine make the categories more precise in e.g. DSM-5? ICD-10? Perhaps, and that is welcome as specificity has always been the aim. But AI and machine learning must account for evolutionary changes in disease. The data base would need continuous updating.
Precision medicine promises to establish more precise classifications for diseases, especially, recently, for malignant tumors, by applying more specific data derived from biogenetics. For example in a recent partnership between a pharmaceutical company and a cancer center, 10,000 cancer patients’ tumors were further analyzed for actionable mutations that would suggest further clinical trials to identify treatment options. When subtypes of tumors were further classified there were too few identified similar tumor subtypes to proceed to a trial. Precision, perhaps? But too many classes. It might be possible for a cooperative study and new classification if recruitment of sufficient number of patients with a sub-subtype were to be identified nationwide or worldwide. Genetic tumor markers, permutations, identify classes of tumors that are not usually recognized in the ICD-10 classification and this may be the role of precision medicine? This in order to treat non-responders (the participants whose tumor markers (are different from the majority) in Phase III clinical trials.
On the contrary medicine promulgates that we ought to treat theperson with the disease and not the data points of a disease, not a mechanical application of data. Observations confined to data-point-generated concepts are blind. There is no disease without a concept, in spite of an array of data-points. Personalized medicine relies on a conceptual model of disease based on physical examination, pathophysiology, imaging, laboratory tests experience, and large clinical trials or meta-analyses of trials.
Without the conventional classification of diseases, e.g. ICD-10, precision medicine would entail innumerable classes of diseases. There would not be a class that contains 1.infectious 2.parasitic, 3.malignant, or 4.degenerative diseases. Each disease subtype would be in a separate class. There might be no class that contains all malignant lymphoma subtypes. Lymphoma now lists more than 60 subclasses.
Precision medicine would consider each patient’s measurements to exhibit a separate distinct disease, a class made up of one patient by collecting innumerable distinctive data points of its own without overlap: no projectability from one disease to a similar disease, nor from one individual to another. This would allow treatment unique to that person’s disease who does not share a class with other patients. But artificial intelligence and machine learning depend upon large data sets, upon population-level data.
Precision medicine treatment is not feasible because medical evidence is based on general tenets, induction, derived from current diagnostic criteria (perpetually undergoing refinements, clinical trials and epidemiology applicable to all or most patients within the disease class.)
Diseases are not stable entities but evolve during treatment to resist treatments, the modal nature is altered by treatment e.g. MRSA, influenza, cancers. the origin of tuberculosis in the middle East is at least 9000 years old1113 Hershkovitz, Israel; Donoghue, Helen D.; Minnikin, David E.; Besra, Gurdyal S.; Lee, Oona Y-C.; Gernaey, Angela M.; Galili, Ehud; Eshed, Vered; Greenblatt, Charles L. (15 October 2008).”Detection and Molecular Characterization of 9000-Year-Old Mycobacterium tuberculosis from a Neolithic Settlement in the Eastern Mediterranean”. PLoS ONE. 3 (10): e3426. and common resistant mutations are numerous, Thr(ACA), Asn(AAC), Ile(ATC), Thr(ACT), Gly(GGC) inhA promoter -15T and evade drug therapy.
Just as coronaviruses have become the current entity to manifest evolution and transmission across species.
Malaria has defied cure based on the point mutations and gene amplifications e.g.crt, dhps, dhfr, mdr1, and even artemisinin resistant malaria mediated by point mutations attributable to  Kelch13.2214Menard D, Fidock DA. Accelerated evolution and spread of multi-drug resistant Plasmodium falciparum takes down the latest first-line antimalarial drugs in southeast Asia. The Lancet, Vol.19, issue 9, p 916-17, 2019 15Zhao Y,Liu Z, Soe MT. et al Genetic variation associated with drug resistance markers in asymptomatic Plasmodium falciparum infection in Myanmar.Genes,Vol.10 Issue 9, p 692, 2019 16 Darwin C. The origin of Species. (Variorum text) Chap 14. P 760.
EBV sequesters in B cells and silently reproduces only to remain viable. Herpes hides in neural ganglia for many years until opportunity arises.
Our taxonomy of disease is based on observations that, given signs and symptoms, diseases reflect the relationship between the disease, its cause and resistance to interventions, natural and therapeutic, and the patient’s pathophysiology; each patient manifests symptoms and signs that mimic, but do not match exactly those in others with the same disease.
Disease classification is a homeostatic cluster of features that distinguish it from related diseases. “We carve nature at its joints.” accordingly.
Leukemia requires a predicate, a sortal predicate if you will, thus ALL, AML, CLL, B-cell lymphomas are categories of the disease, yet antigenic markers distinguish among these diseases to allow targeted therapy, e.g. based on cytogenetics and therapeutically on molecular markers KIT, FLT3-ITD, NPM1, CEBPA, PD-L-1, CD-19 perhaps CD30.
As our nominative classification of these diseases is based on testing for the antigen expressions, and as therapy induces changes in the cellular expressions, so tumor cells become resistant to further elimination by transmutations to acquire resistance; surviving clones in minimal residual disease (MRD.) Discrete classes, carved at their joints, provide more specific treatments. Or so it is hoped. The mechanistic approach of precision medicine is here applicable. But the measurements required to perform to characterize a condition are on a asymptotic curve.
Evolutionary changes in the body, dependent on both external and internal environments, allow diseases to escape the immune system as illustrated above. Tumor cells evolve to become refractory to chemotherapy. Loss of responsiveness to treatment with monoclonal antibodies (mAbs) such as
rituximab is a serious complication during therapy of B-cell malignancies but the mechanisms responsible for it are not well understood. Even CAR-T therapy directed against CD-19 has not shown success in all cases of leukemia and lymphoma; CD30 is a promising alternative antigen target. BTK inhibitors such as ibrutinid may play a similar role in some
non-Hodgkin’s lymphomas.
Clones of tumor stem cells evolve to resist intervention.33
Molecular evolution, even in laboratory-controlled constant supporting conditions are determined by stochastic processes. Escherichia coli mutation and was measured over 60,000 generations showing only a slight decline in fitness; and these changes were balanced against multiple beneficial variants.4417.Good B.H., MaDonadl M.J. et al. The dynamics of molecular evolution over 60,000 generations. Nature 551, 2 November 2017,;4550
Thus, generalizability (induction) is forfeited to particularity, sensitivity to specificity, and randomized controlled trials would suffer from lack of participation since each disease would be treated somewhat differently, depending on the definition based on the character of the disease, genetics, as well as on the sufferer. Thus, precision medicine would usurp the medical treatment available based on randomized controlled trials and would evolve into ‘customized’ treatment. Treatment protocols would be suited to the genomic markers of the disease and these degrees of difference might require different therapies. This would eventually and controversially obviate the need for randomized trials
Research options :
In place of RCTs precision clinical trials might include n of 1, or even master protocols such as umbrella and basket trials. One SNP does not make an entirely new disease. That is the foundation of umbrella protocols, Tumor cells regulate antigenic self-expression in order to survive the treatment by protocol drugs, and they usually succeed. Even ALL is not vanquished, thus CAR-T Therapy. Yet which tumor cells to search? which fluid biopsies of CTCs or ct-DNA to sort and harvest and insert a chimeric antigen?