|   |   |
  | Course of LABORATORY MEDICINE Principles of clinical reasoning and the problem of cancer       Clinical reasoning is a branch of logical problem solving, and follows its own rules. As discussed in the lecture on nosography, diagnosis is the procedure or assigning the condition of a patient to a group. This assignement is usually probabilistic and plagued by possible errors. Clinical reasoning provides a method that aims to minimize errors and to suggest a logical path that allows their correction.       The study of a patient starts with the collection of his or her anamnesis, and his or her physical examination. These steps are usually described in the courses of medical semeiotics. One should resist the temptation to establish the diagnosis at these early steps of evaluation       The next steps of clinical investigation are usually a standard blood and urine test, and possibly, if suggested by the previous step, some imaging investigation (e.g. a chest X-ray film). The scope of the standard blood and urine tests is to evaluate the function of as many organs as possible; the anamnesis and physical examination may suggest specific laboratory tests to be added to the standard ones.       Once a sufficient amount of clinical information on the function of different organs has been collected, the physician will draw a list of the possible diseases that may explain the clinical findings. As discussed in the lecture on the standard blood test, it is useful to draw this list considering separately each organ and each function, and taking into account that every organ plays more than a single function and that every function requires the cooperation of several organs. The goal of this step is to construct a comprehensive list of possible diagnoses; some will be more likely than others, but statistical considerations are better deferred to a later step of the diagnostic procedure, because of two main reasons: (i) at this stage many possibilities will be reasonably likely (see the lecture on statistical considerations, and the discussion of Bayesian statistics); (ii) on large numbers of cases even unlikely possibilities do sometimes occur, thus a physician who in his or her career will see a large number of patients should never dismiss a diagnostic hypothesis that might occasionally turn true.       At this point the physician will consider every item of the list and prescribe the appropriate laboratory analysis or test to include or exclude its presence. Hopefully, at the end of this step several of the original hypotheses will have been excluded and a likely diagnosis can be proposed. In a young patient the coexistence of two diseases is uncommon; thus your hypothesis should tend to explain the whole clinical picture; by contrast in elderly patients the coexistence of more diseases is relatively common, thus posing two diagnoses (or more) is usually justified. Some examples are reported below; the student may also take advantage of the clinical tutorials provided in this website or in any other specialized website (e.g. Medscape).       Grand diseases and incomplete diseases. This is a concept due to the great French neurologist Jean Martin Charcot. Charcot collected all symptoms and signs of the same disease from his casistics of many patients. This collection he called the "grand disease". Every single patient presented only a fraction of the possible symptoms and signs of the disease; Charcot called these the "incomplete" or "little" diseases. Charcot's concept maintains its validity. There are many reasons why the list of possible symptoms and laboratory signs of every single disease is much longer tham the list of symptoms affecting every single patient; for example, we may see the patient in an initial stage of the disease, when all symptoms and signs have not yet developed (e.g. a patient suffering of liver cirrhosis may present jaundice, and increased liver enzymes, but not coagulation defects, hypoalbuminemia, and ammonemia); or the disease itself may be milder than usual and develop only part of its damginng potential (e.g. damage of the adrenal glands or the hypophysis may produce a variant of Addison's disease in which secretion of glucocorticoid hormones is more impaired than that of aldosterone). Since every patient may present and incomplete clinical picture absence of a sign or symptom does not rule out the diagnosis.       Can a computer be a physician? Many computer programs have been developed to help the physician in the choice of the correct diagnosis and treatment. These are at present quite unsatisfactory, but the progress in the field is rapid and it is to be expected that they will be rapidly improved. Moreover they constitute a useful learning instrument. For example you can play with symptomate, which has been designed for patients, but has quite a good constructed interface; or you can try diagnosaurus which is part of the suite Access Medicine. Diagnosaururs is not strictily speaking a program for diagnosis; it is a simple relational database that receives a symptom as the input and outputs all(?) diseases in which that symptom may appear. To understand the possible role and utility of these programs, we can consider the following. 1) Systematic pathology is strictly linked to nosography: these disciplines define which diseases affect humans, which are their symptoms, laboratory findings, etc. No medicine exists outside systematic pathology. Nosography is the answer to the question "which are human diseases?"; systematic pathology answers the question "which are the symptoms and signs of each disease?", A properly programmed computer can be proficient in these disciplines, which ultimately are essentially an ordered collection of empirical data. 2) Diagnosis requires clinical reasoning, which goes backward with respect to systematic pathology. Diagnosis starts from the empirical data collected via the anamnesis, physical examination, laboratory tests and other instrumental investigations, and answers the question: "which disease affetcs this patient?". The physician must elaborate a list of possible explanations for the symptoms and signs presented by the patient, and then exclude or confirm each hypothesis using the appropriate secific test. A computer can carry out this task quite effectivey, as it is essentially a problem of data mining from an extensive relational database 3) A computer is quite poor in the process of data collection. Data collection requires the collaboration of the patient, which can only be obtained within the setting of patient-physician relationship. The patient will reveal his or her anamnesis, and accept the manipulation involved in the physical examination and blood samples collection, and will collaborate only with the empathical and caring physician. Empathy is crucial in the therapeutical relationship. Moreover the physician may notice facts that escaped the attention of the patient, e.g skin discoloration, asymmetry in gait or standing, etc. The computer might become in the future an important instrument at the service of the physician, not a replacement of the physician. Learning how to use diagnostic programs may be a very good investment for the physician, provided that a specifically designed program is selected (chatgpt and similar AI based programs are NOT specifically designed for this use and should not be considered). Legal and professional responsibility will always rest on the physician, not on the computer!.       Clinical example 1: main complaint is blood in the sputum since several weeks. Familiar, physiological and pathological anamnesis are negative; physical examination reveals moderate fever and an area of obtuse sound at percussion and some wheezes. Laboratory findings: leukocytosis, mild anemia, high ESR and C reactive protein. Which clinical conditions may cause the above picture? Consider: bronchopneumonia; tuberculosis; chronic obstructive pulmonary disease; lung or bronchial cancer; left heart failure; pneumoconiosis. The first items in this list are also the more likely. Which laboratory or other tests are most appropriate for each of the above conditions? Bronchopneumonia: chest X-ray; microscopic and cultural examination of the sputum; search for antibodies in the serum. Tuberculosis: same as for bronchopneumonia Lung or bronchial cancer: chest X-ray; cytological examination of the sputum; bronchoscopy with biopsy Left heart failure: measure ejection fraction and central venous pressure; dynamic echocardiography; ECG Peumoconiosis: chest X-ray; functional ventilatory tests; hemogas analysis       Many of the above conditions indicate the same clinical investigation (e.g. chest X-ray film), but you will carry out all of them, or at least as many as reasonably feasible, taking into account that some are invasive, thus should be delayed till non invasive ones suggest their necessity: also some analyses may have contraindications (e.g. bronchoscopy in the presence of an open infected lesion may favor the spread of bacteria to previously unaffected areas of the lung).       Clinical example 2: 80 years old male; weakness since several months; more recently cutaneous petechiae. Anamnesis negative; physical findings negative except for the petechiae. Mild tachycardia. Laboratory findings: normocytic anemia; moderate thrombocytopenia. Which clinical conditions may cause the above picture? Consider: chronic internal hemorrhage (usually from the GI tract); vitamin or other nutritional deficiency; bone marrow aplasia in its initial phase; leukemia; lymphoma; allergic reactions that destroy red cells and platelets. Which laboratory or other tests are most appropriate for each of the above conditions? Chronic internal hemorrage: search for occult blood in the feces; count reticulocytes; measure erythrocyte lifespan. Malnutrition: Measure the serum concentration of iron, transferrin, vitamins (folate, ascorbic acid, etc.) Bone marrow aplasia, leukemias and lymphomas: bone marrow biopsy.       Clinical example 3: 35 years old patient complaining of sudden pain at a major joint (e.g. elbow). Pain increases with movement. Anamnesis mostly negative, but paternal grandfather suffered of gout. Physical examination reveals local reddening and tumescence, and excludes superficial lesions that may be erroneously referred to the joint (e.g. wounds, insect bites). Which clinical conditions may cause joint pain? Consider: traumas, especially sport-related traumas; hemophilia (hemorragic arthritis); acute arthritis. Acute arthritis may be due to: autoimmune diseases; gout (present in the familiar anamnesis); infection (septic arthritis). Which laboratory or other tests are most appropriate for each of the above conditions, and may be added to the standard blood tests? Traumas are disclosed at the anamnesis; an X-ray film may reveal bone lesions; tendineous lesions are usually silent at the X-ray, but visible by NMR. The possibility of hemorragic arthritis indicates measurement of PTT, PT, and/or INR. Autoimmune arthritis usually occurs in the context of other autoimmune lesions (e.g. dermatitis, glomerulonephritis). Prescribe a search of autoantibodies. Gout arthritis is diagnosed after finding increased urate concentration in the blood and urine. Septic arthritis does not indicate specific tests, but in the standard blood test one will observe leukocytosis, with major increase of granulocytes. In all cases of arthritis (except hemorragic arthitis) the standard blood test will be positive for generic signs of inflammation (increased erythrosedimentation rate; increased reactive C protein).       The problem of cancer. Cancer is a complex type of disease in which a cell clone loses its mechanisms of regulation, and reproduces itself continuously invading the organism. Cancer cells may invade the limphatic or blood vessels and be disseminated to organs different from those in which the cancer has originated producing secondary cell masses called metastases (secondary cancers). The prognosis is poor because of the functional destruction of the organ(s) that have been invaded by cancer cells.       Cancer may cause and explain essentially every conceivable collection of laboratory and physical findings; thus cancer will figure in essentially every list of diagnostic hypotheses. In the case of complex clinical cases it is tempting to postulate a cancer diagnosis, multiorgan and multifunction failures resulting from metastases. Cancer is a frequent cause of logical pitfalls in clinical reasoning, because it tends to overshadow other possible hypotheses; the take home message is: feel free to include one or more cancers in your list of diagnostic hypotheses, but avoid to exclude other hypotheses just because cancer seems to fit so well.       Essentially all organs and tissues may originate a (primary) cancer, but the frequencies of cancer is highly variable among different organs; frequent cancers are those orginated in the following organs: - colon and rectum - lung and bronchi - uterus - breast (in women; very uncommon in men) - prostate - pancreas - skin (melanomas, squamous carcinomas) - hematopoietic system (leukemias, lymphomas)       In some organs primary cancers are common only if there are preexisting lesions: e.g. in the liver hepatocellular carcinoma is strongly favored by such chronic lesions as cirrhosis or chronic viral hepatitis.       Laboratory diagnosis of cancer is difficult because cancer has essentially no specific and fixed laboratory signatures, and may mimic every organ insufficiency or failure. Moreover, laboratory signs of cancer related organ failure usually occur late in the development of the disease, whereas early diagnosis (when laboratory signs are scanty) is essential for prompt therapy. Some examples are as follows: - pancreas cancer may cause a type I diabetes, because of the destruction of the Langerhans islets - leukemias and lymphomas may cause anemia thrombocytopenia or pancytopenia because of the invasion of the bone marrow - liver cancer (primary or secondary) may cause hyperbilirubinemia - colon cancer may cause anemia (because of an unnoticed chronic hemorrage). - cancers of endocrine glands may produce the hormones of the gland       Biochemical markers of cancer are proteins that can be released by cancer cells. They are usually searched in the blood, possibly also in the urine. Tumor-Specific Antigens (TSA) are produced only by tumor cells; Tumor-Associated Antigens (TAA) are produced also by normal cells, but cancers produce them at higher levels. Since TAA are also produced by the normal cells, their analysis may have significant numbers of false positives and false negatives. Examples are: - Alphafetoprotein, in liver cancers - Carcino-embrionic antigen, in bowel tumors - PSA: the prostatic antigen, increased in prostate cancers - mesothelin is a protein overexpressed by tumors of mesothelial origin - claudin is a protein overexpressed by tumors of the GI tract and the lung ...       Paraneoplastic syndromes and their markers. A cancer cell possesses the entire human genome, and since its gene expression is disregulated, it may sometimes produce and secrete proteins and hormones that are not characteristic of the tissue from which the cancer originated. Moreover, the immune response against cancer may cause organ damage. All these adverse effects of cancer go under the broad category of the paraneoplastic syndromes. Some examples are: - fever; weight loss; fatigue; ... - neurological symptoms: seizures, dizziness, paresthesias, ... - endocrine disorders: Cushing-like syndrome; hypercalcemia - rheumatologic symptoms: diffuse arthritis, joint pain - cutaneous symptoms: itching, flushing - glomerulonephritis       Diagnosis of cancer usually requires a biopsy!; imaging methods may provide strong diagnostic clues; laboratory data in this case are accessory and may indicate the need of imaging investigations (e.g. NMR total body; colonscopy; etc.). Never mention this diagnosis to the patient till a biopsy has been performed!       Home of this course |
  |