CHEMICAL CARCINOGENESIS (JAMA 266: 681, 1991; Science 250: 1644, 1990; Science 251: 10 & 387, 1991; little has changed since and this is no longer "cutting edge"; update for the truly-hardcore Mut. Res. 489: 17, 2001)

Classic carcinogenesis experiments disclosed that key steps are often the induction and promotion of cancer by chemicals. Chemical carcinogens and human health.

The historic Delaney Clause from the 1950's forbade the presence of any "cancer-producing chemical" in any concentration in U.S. food. The selective enforcement of this unrealistic (obsolete, frankly silly nowadays) ideal was highly political and kept lawyers busy: Nature 358: 181, 1992.

The Delaney Clause was finally modified in 1996 to a "no harm" requirement. (Thanks for once, Newt.) But the whole business of government regulation of what can and cannot be in food remains arcane.

Now is a good time to learn the following associations:

Soot: Cancer of the scrotum ("chimney sweep's cancer" -- * discovered by Percival Pott)

Cancer chemoRx: Acute leukemia (*  the bad ones include cyclophosphamide, chlorambucil, busulfan, melphalan, others -- the alkylating agents)

Cyclophosphamide: Transitional epithelial (mostly bladder) cancers

Other alkylaters: Many cancers (remember nitrogen mustard, bischloromethyl ether, benzyl chloride)

Polycyclic hydrocarbons:: Tobacco smoking-related cancers (lung, larynx, mouth, throat, esophagus, pancreas, bladder, kidney -- * remember 3-methylcholanthrene, benz(a)anthracene and benzo(a)pyrene).

Azo dyes: Bladder cancer (dye factory workers, ?? red-M&M eaters, etc., etc. -- remember "butter yellow" in margarine, "scarlet red" in maraschino cherries, and beta-naphthylamine). Azo dye workers still have a tremendous increase in urothelial cancer, even many years after exposure: Cancer 76: 1445, 1995; also J. Occup. Env. Med. 42: 762, 2000; Am. J. Ind. Med. 46: 505, 2004.

Aflatoxin: Eaters of moldy grain and peanuts (hepatocellular carcinoma, endemic in Africa; the mold is aspergillus species); the region's farmers are now taking effective measures to control it (Lancet 365: 1950, 2005).

Betel nut: Mouth and throat cancer (addictive substance chewed in India and elsewhere; despite older reports the betel nut itself is probably a carcinogen: Lancet Onc. 4: 587, 2003).

* Maté : Uruguayan herbal concoction; with black tobacco, takes blame for Uruguayan epidemic of esophageal and bladder cancer (Cancer 67: 536, 1991; Cancer Ep. 12: 508, 2003.)

* Safrole: Sassafras (stomach cancer? liver cancer? other cancers?; a free-radical generator Science News 114: 109, 1993) and binds to DNA (CMAJ 162: 359, 2000). Despite warnings from the FDA, politics still allows the tea to be peddled extensively as a "complementary holistic remedy".

Aristolochia: A "holistic" herbal remedy that has caused end-stage renal disease in hundreds of people, many of whom went on to develop transitional cell kidney and bladder cancers (J. Ethnopharmacology 94: 245, 2004; Food Chem. Tox. 41: 29, 2003).

Vinyl chloride: Angiosarcoma of the liver (factory workers)

Chromium, nickel: Lung cancer (factory workers -- scramble chromosomes and somehow enhance the effectiveness of real mutagens: Tox. Let. 127: 63, 2002). Cr⁺⁶ is the worst known mutagen among the metals.

Cadmium: Lung cancer (strong link) and prostate cancer (weaker; battery factory workers)

Asbestos: Lung cancer, mesothelioma (scrambles chromosomes)

* Silica / Sand: Lung cancer. Yes, sand is now an "official" EPA carcinogen (Am. J. Epidemiol. 153: 695, 2001 shares my skepticism). The numbers are very soft and there is no easy-to-believe mechanism.

Arsenic: Arsenic given as a medication for syphilis was famous for causing skin cancers (amplifies genes -- see Science 241: 79, 1988). Some arsenic occurs naturally in ground water, and this has been a major concern recently. Especially, bladder cancers might have something to do with arsenic in US drinking water (Am. J. Epidem. 153: 411, 2001).

The business about arsenic in drinking water in the US is extremely politicized, which is easy to do in the absence of hard scientific data: Science 291: 2533, 2001; CA 51: 254, 2001; CMAJ 166: 69, 2002.

By contrast with the questionable risk in the US, there is no question that the epidemic of arsenic exposure in Bangladesh due to deep wells is increasing cancer rates there (Am. J. Pub. Health 94: 741, 2004).

PCB's: Polychlorinated biphenyls (pollutants, perennially accused of causing human cancers; hard evidence, i.e., increased cancer in people who actually WORK with the stuff and are heavily exposed, is conspicuous by its absence, which must be why only the World Health Organization is still writing about them (J. Tox. 40: 457, 2002)

Saccharin: Bladder cancer (in huge doses given to animals, but epidemiologically not a significant risk to human users). Saccharin was banned in the 1970's for political reasons.

Cyclamates: Same story as saccharin.

* We'll talk later about the obviously-false aspartame claims, the work of a single internet activist who has avoided civil and criminal penalties only because his/her true identity has never been discovered.

Human feces: Several known carcinogens, including those derived from bile salts (*  try and ban feces, Senator Delaney!)

Benzene: Leukemias and related problems.

* Vitamin C can decarboxylate sodium benzoate in warm, light-exposed soda pop, increasing levels above what is considered "safe" for drinking water. This occasionally gets government and media attention, though it's never more than a very few lots that are found to be tainted.

Phenacetin: Transitional epithelial (mostly bladder) cancers

Anabolic steroids: Liver cancer (this particular risk is relatively small; more about "roids" later)

Estrogen: Endometrial hyperplasias and carcinomas

Ferric ion: Liver cancer (hemochromatosis patients); perhaps many other cancers ("free radical generator")

* Herbicides: Chlorophenol herbicides occasionally show small statistical links to some of the soft tissue sarcomas, which are fairly rare anyway: Epidemiology 10: 788, 1999; Am. J. Epidem. 145: 1061, 1997; the numbers aren't very impressive; this has been tossed around for years.

Some environmental carcinogens are direct-acting ("activation-independent"), and exert their effect directly. However, the majority (procarcinogens) require metabolic conversion (activation) to produce carcinogenic molecules (ultimate carcinogens).

Famous direct-acting carcinogens include the alkylating agents (cancer chemotherapeutic agents) and a few acylators. Some heavy metals actually depolymerize DNA.

All the others, including polycyclic hydrocarbons (smoke), aromatic amines, amides, and azo dyes, natural plant products, and nitrosamines all require activation to ultimate carcinogens. Often (but not always) the carcinogen is activated by the hepatic P-450 mixed function oxidase system.

Probably all chemicals that really induce cancer are mutagens ("genotoxic carcinogens"). You don't want any more exposure to these than absolutely necessary.

The non-mutagens ("non-genotoxic carcinogens") act by promoting cell division ("promoters"); these are clearly dose-dependent and the effect is reversible when the promoter is eliminated. These are a lot less dangerous and include lots of common substances -- hence the absurdity of banning "all traces of anything that causes cancer".

A rule that works most of the time is that the actual carcinogen either damages DNA directly (the alkylating and acylating agents) or is a potent electrophile (*  the epoxide ultimate carcinogens derived from polycyclic hydrocarbons, vinyl chloride, and aflatoxins; the N-hydroxylated dye metabolites; the alkyldiazonium ions derived from nitrosamines, etc., etc. etc.: best review article is still Cancer 47: 2327, 1981.)

Review of how environmental carcinogens produce mutations: JAMA 266: 681, 1991 (still good).

When (not "if") you, the physician are asked about media and government claims that something causes cancer, please bear in mind that the relationships that have held up have been striking, apply to animals too, and make sense biologically.

Where the link has proved genuine:

• the relationship is striking, and not just the result of selective memory (i.e., hype any "cause of cancer", and cancer patients and their families, seeking to know "why me?", will remember being exposed to more of it than do controls; real relationships hold up in prospective studies);
• there is a dose-response curve that makes sense (as you read more and more "scare articles", you'll be amazed how seldom this is present)
• there is a good animal model, and not just huge amounts of some substance that are probably acting as a promoter (see below);



• it makes sense biologically (i.e., you are dealing with something you can demonstrate to be a mutagen / something that interacts with DNA).

The most dubious "carcinogen" in the public eye in the 1990's was high-tension electric lines. Only in deeply flawed "epidemiologic studies" have "statistical risks" been identified. There is no theoretical mechanism, no one has been able to induce cancer in animals this way, fields orders of magnitude higher have no apparent effect on bio-molecules or cells, and the "electromagnetic field exposure" from the body's own beating heart is far greater. The newer epidemiologic studies haven't shown an effect, either (Br. Med. J. 307: 895, 1993). See also JAMA 265: 1438, 1991; Cancer 68: 455, 1991; Pediatrics 88: 630, 1991; Br. Med. J. 313: 1047, 1996; NEJM 337: 1, 1997. The major article claiming a link was finally branded a fake by Br. Med. J. 319: 337, 1999. The business seems to have ended. (The "cellular phone" business was even sillier. Ask a tort lawyer.)

* In June 2005 there was a pronouncement by the EPA that they were gravely concerned about the safety of teflon, one of the most inert substances in existence, because a chemical used in its manufacture was a carcinogen. My search of the NIH database did not show a single publication supporting this claim.

Agent Orange was contaminated by the experimental carcinogen 2,3,7,8-tetrachlorodibenzoparadioxin (TCDD), which remains present in measurable quantities in some veterans even now (Am. J. Ind. Med. 30: 647, 1996). Despite the decision by politicians to compensate Vietnam veterans with lymphoma (the son of Admiral Zumwaldt, who ordered the spraying of agent orange, got lymphoma...) and (1993) tobacco-induced lung cancer (I'm not making this up), any link between agent orange (dioxin) and a plethora of alleged health problems (cancers, birth defects) remains very soft. So is evidence that most of our soldiers were even exposed. See JAMA 265: 898, 1991; Am. J. Pub. Health 81: 289 & 344, 1991; Arch. Env. Health 53: 199, 1998 (Air Force; no chloracne or noted increase in common acne in veterans who sprayed it during Operation Ranch Hand); Am. J. Epidem. 148: 786, 1998 (no increased mortality; no increase in total cancer); J. Occ. Env. Med. 39: 740, 1997 (VA study finds lung cancer claim fails totally); Arch. Env. Health 51: 368, 1996 (gestational trophoblastic disease claim fails); Epidemiology 7: 454, 1996; Ann. Epidem. 5: 414, 1995 (VA; Hodgkin's claim fails completely); Epidemiology 6: 17, 1995 (claims of more stillbirths and birth defects fails completely). No link to prostate cancer: J. Urol. 166: 100, 2001. No link to trophoblastic disease in the Vietnamese people: Arch. Env. Health 41: 368, 1996. People heavily exposed in industry have only a slight increase in overall cancer risk, even assuming that the effect isn't due to confounding variables (Occ. Env. Med. 53: 606, 1996; Env. Health Perspect. 106 S2: 663, 1998); one epidemiologist actually showed how to juggle the statistics, including studies the EPA chose to ignore, to claim dioxin protects against cancer (sort of like broccoli sprouts, I guess; Reg. Tox. Pharm. 26: 134, 1997.) Nevertheless, in 1994, the Environmental Protection Agency issued a report concluding that dioxin as among the "greatest threat[s] to public health", i.e., was a grave danger that could be the cause of 1.3 out of every 100 American cancer deaths. Of course, its own Science Advisory Board refused to accept this groundless claim, both in 1995 and after the EPA's 2000 revision (Tox. Sci. 64: 7, 2001 points out even the EPA is not allowed to divide by zero; also Reg. Tox. 36: 211, 2002, which considers among other strange things the EPA's willingness to believe in "U-shaped dose-response curves"). In 1997 the government decided to compensate Vietnam vets whose children have neural tube defects; again this is probably politics rather than science. Of course, the Hanoi government claims a tremendous increase in birth defects "caused by Agent Orange"; there was a conference in 2001 (Nature 413: 442, 2001) that produced the expected agreement for joint study (Nature 416: 252, 2002). The fiasco continues: One group estimates "dioxin body burden" in survivors of "Operation Ranch Hand", and discovers that the "most heavily exposed" people have 50% LESS cancer than controls (Exotoxicity and Env. Saf. 50: 167, 2001); they come up with a weird explanation ("both a promoter blocker and a cancer causation agent...") for this.

* John Travolta's "A Civil Action" popularized the idea that the Woburn cluster of childhood leukemia cases resulted from trichloroethylene from a tannery entering the drinking water. I could not find the details of the case, but trichloroethylene-exposed workers are not coming down with more leukemia themselves (Epidemiology 9: 424, 1998; Int. J. Occup. Med. 11: 81, 1998; Cancer Causes & Control 8: 406, 1997). Whatever really happened; I have thought that perhaps trichlorophenol (which has been used in tanning and which is an animal carcinogen) is more suspect.

RADIATION CARCINOGENESIS

Exposure to high-energy photons (ultraviolet, ionizing radiation) is well-known as a cause of cancer.

From 1928-1955, "Thorotrast", a complex of thorium dioxide and a carrier dextran, was used to image the livers and spleens of a few million humans. This was monumentally stupid. The long-lived isotope stays in the body, emitting radiation, for the rest of your life, and cancers followed years or decades later (J. Tox. 35: 199, 1997).

Atomic bomb survivors (Japan, Marshall Islands) have greatly increased incidences of all the common leukemias (except CLL; the incubation time is a few years), and minor increases in many (but not most) solid tumors (remember thyroid, breast, salivary gland, lung).

Chernobyl's children (thyroid cancer from radioactive iodine, other problems have been less prominent): Nature 359: 21, 1992; update Lancet 358: 1965, 2001.

Interestingly, children conceived after their parents were exposed to radiation at Hiroshima and Nagasaki do not exhibit any measurable increase in any identified health problem so far. This "non-news" is very important scientifically (Am. J. Hum. Genet. 52: April 1993).

On the average, 14% of your annual radiation exposure is from your diagnositic x-rays. The cancer risk that this poses is unknown.

A group in 2004 made the dubious assumptions that (1) there is a linear relationship with the atomic bomb survivors, and (2) countries that do more x-rays don't find more cancers. They estimated the diagnostic x-rays cause 700 extra cancer deaths annually in the UK (Lancet 363: 344, 2004).

More about the "radon in your energy-efficient home causes lung cancer" business later (in any case, it's no measurable risk for anything else important; Lancet 341: 1127, 1993; more Lancet 355: 1888, 2000). Since the early 1990's there has been amazingly little work on radon despite the ongoing political stuff. Nobody's shown an increased risk from living near nuclear power plants (JAMA 265: 1438, 1991).

Occupational radiation carcinogenesis:

Old-time radiologists who tested their fluoroscopes using their own hands developed lots of leukemias, Hodgkin's disease, and skin cancer.

Radium paint workers who put their brushes in their mouths developed bone and nose cancers.

Uranium miners have a greatly increased incidence of lung cancer, supposedly even if they do not smoke.

* In the US, the worst harm was done to Navajo miners. Historians see Am. J. Pub. Health 92: 1410, 2002. New series: Cancer 89: 2613, 2000 (almost all the miners with cancer smoked.)

Iatrogenic radiation carcinogenesis:

People given high doses of radiation for ankylosing spondylitis (x-rays) or polycythemia vera (radiophosphorus) have greatly increased incidences of all the common leukemias (*  except CLL).

Patients treated with radiation therapy for acne (!) develop multiple skin cancers.

Newborns treated for mythical "enlarged thymus" developed many thyroid cancers as young adults.

Ultraviolet radiation is the principal risk factor in most skin cancers (basal cell, squamous cell, malignant melanoma).

Suntanning offers only modest protection from the wavelengths that cause cancer and elastosis ("aging of the skin").

Radiation appears to initiate cancer just as chemical carcinogens do -- by causing mutations.

* The trademark ultraviolet light mutation is CC->TT (Proc. Nat. Acad. Sci. 90: 4216, 1993).

ONCOGENIC VIRUSES

Viral (RNA, DNA) causation of cancer is well-documented in the lab, and is important in some (but probably not most) human cancers.

Polyoma virus and SV40 are linked to a variety of animal tumors.

* There is perennial discussion of whether exposure to SV40 in contaminated polio vaccines has caused an increase in the rate of all cancers, or any particular cancer. When people exposed to the virus get certain of the less-common cancers (ependymomas, mesotheliomas, osteosarcomas), the cancer cells express the virus, but if there is any increased risk among these people, it is relatively small (Anticancer Res. 19(3B): 2173, 1999), and numbers still differ tremendously from various centers. Most recently, it seems that SV40 is often expressed in particular cancers, but even after 25 years of worrying, cause-and-effect is hard to show (Am. J. Med. 114: 675, 2003). Intersetingly, even the authors of the last paper make no attempt to show that the involved cancers are more common in recipients of the tainted polio vaccine batches.

Mouse mammary tumor virus is transmitted from mother to child in the milk.

Feline leukemia virus causes a contagious leukemia in cats.

Closer to home: Wart virus ("human papilloma virus", HPV) causes warts ("benign tumors") in humans, and certain strains cause cancer of the uterine cervix, vulva, and penis in humans (Nature 336: 765, 1988 was the breakthrough article).

* For your future reference: HPV oncogenic protein E7 inactivates the product of tumor-suppressor gene RB, while E6 inactivates tumor-suppressor gene p53 product and prevents it from repairing damaged DNA (Proc. Nat. Acad. Sci. 90: 3988, 1993; Proc. Nat. Acad. Sci. 91: 2436, 1994. These two genes do the transformation in mice: Cancer Res. 52: 4420, 1992).

Epstein-Barr virus ("infectious mononucleosis virus") is necessary (but not sufficient) to cause African Burkitt's lymphoma, and is etiologic in Chinese nasopharyngeal cancer, immunoblastic lymphoma, and * Eskimo endemic salivary gland adenocarcinoma.

* Using Epstein-Barr as a tumor marker for prognosticating Chinese nasopharyngeal carcinoma: NEJM 350: 2461, 2004.

Hepatitis B virus has been known for decades to be a major cause of hepatocellular carcinoma. It became clear in the 1990's that hepatitis C virus is also important (NEJM 325: 675 & 729, 1991; lots more). They probably effect this by acting as mitogens, allowing special opportunities for genetic damage (PNAS 86: 8852, 1989; PNAS 87: 6791, 1990; Cancer Res. 51: 1278, 1991). Hepatitis B (and perhaps C) also inserts itself randomly in the genome, with a range of possible effects (Ult. Path. 25: 497, 2001).

HTLV-I causes epidemic leukemia in Japanese humans.

* Adenovirus, E2F protein and the Rb gene product: Nature 358: 181, 1992; Science 258: 424, 1992.

Despite all this, the common human cancers (except as noted) do not seem to be contagious. Viral carcinogenesis promises to be an area of continuing interest.

OTHER REPUTED CARCINOGENS

Foreign-body carcinogenesis and carcinogenesis by repeated trauma are possibilities that worry patients.

There is almost nothing to suggest that foreign implants cause cancer. (The breast implant hype and fiasco: NEJM 326: 1649, 1992.) Cancers caused by bile duct flukes or schistosome eggs probably are due to their effects as promoters. Joint replacement doesn't cause cancer at the joint or elsewhere: Cancer 94: 3057, 2002.

The weight of evidence is that mechanical injuries do not cause cancer, though this is often alleged in lawsuits.

The only likely exception is fibromatosis / fibrosarcoma where there was massive soft-tissue injury (AMFJP 19: 152, 1998; also Surg. Gyn. Ob. 169; 104, 1989; J. Ped. Surg. 34: 1130, 1999) Both make sense -- cells that do not ordinarily divide will have divided in response to the trauma, letting Nowell's Law operate and producing a tumor with the phenotype of the dividing cell.

* The traditional wisdom is that head trauma places people at risk for meningiomas, for some unknown reason; not surprisingly, the recent studies show little if any effect (Cancer Causes & Control 9; 109, 1998; Int. J. Epidem. 27: 579, 1998).

Marjolin's ulcer is well-differentiated squamous cell carcinoma arising in skin that has been badly traumatized (i.e., burn, chronic ulcer), where ongonig regeneration has provided the opportunity for mutated cells to overgrow.

There is a great deal of speculation and anecdotal evidence connecting carcinogenesis and prognosis of cancers to mental attitudes.

So far, the best work has failed to support the connection made by folklore, and nowadays real scientists have pretty much stopped examining these claims. The "stress and breast cancer recurrence" claims (Br. Med. J. 304: 1295, 1992; Br. Med. J. 304: 1078, 1992) were finally laid to rest by a huge study (Br. Med. J. 324: 1420, 2002). "Psychic vulnerability" is not a risk for cancer either (Cancer 94: 3299, 2002, twenty-year prospective study; yes really). "Fighting spirit" does not correlate with survival, but being depressed does; I suspect that the latter effect means the cancer has gotten farther and is causing an organic depression (Lancet 354: 138, 1999).

But the relationship between mind and body is clearly a very potent one, and the whole field cries out for more study.

ONCOGENES

Cancer genetics update: Nat. Genet. 33S: 238, 2003.

Oncogenes are DNA sequences within eukaryotic cells that seem to be involved in the development and maintenance of tumors. These genes direct the synthesis of proteins that under some conditions transform a benign host cell into a cancer cell.

Oncogenes are slightly altered forms of proto-oncogenes ("mitogenes") that are essential genes that govern normal tissue growth, differentiation, and apoptosis.

By now, about 200 different proto-oncogenes are known.

Many proto-oncogenes are the genes for hormone or vitamin receptors or the proteins to which they talk, while others seem to be general turn-ons.

When a proto-oncogene is altered to become an oncogene, we speak of its being activated. This is by one of three mechanisms:

• Point mutation: The gene product is stuck on the "on" position.
• Translocation: The gene product is under the control of the wrong promoter, or a fusion gene (half-one gene, half-another) produces a product that promotes cell division.

• Amplification: There are too many copies of the gene. Sometimes there are so many copies that "double minute" chromosomes are formed from them

* As key genes for cellular function, growth, and differentiation, proto-oncogenes have been highly conserved through evolution.

All the families of proto-oncogenes that have been studied so far exist in all vertebrates, producing nearly-identical proteins.

In fact, most of them exist in very similar forms in fruit flies, yeast, etc., etc., though their functions in these species may be different from their functions in vertebrates.

Inconsequential differences in codon sequences (i.e., redundant third bases, occasional amino acids differences), when compared, have so far all yielded the same "phylogenetic tree" as classical comparative anatomy. Actually, this has proved true of all proteins so far studied, and this fact is the strongest evidence I know for the common ancestry (rather than just "intelligent design") of all living things. If this were NOT true, Darwin's macroevolution would be refuted. Clonal selection in Darwin's world: Nature 363: 208, 1993.

Oncogenes were originally discovered in transforming retroviruses ("the RNA tumor viruses").

Retroviruses are ubiquitous, generally harmless RNA viruses (HIV is obviously an exception). The RNA code is transcribed onto DNA, which is then integrated into the host genome. "Viral oncogenes" turned out to be cancer-producing genes that the viruses had just happened to pick up ("transduced") while growing in established tumors. In fact, the proto-oncogenes were originally discovered as counterparts in the normal genome to the deadly viral oncogenes.

Typically a "viral oncogene" is a proto-oncogene minus its regulatory sequences, or with a characteristic mutation, or in an excessive number of copies ("amplification"). They are capable of causing cancer by themselves, and hence are very different from their normal counterparts (i.e., have been damaged several times).

As we have noted, a proto-oncogene that has acquired the ability to cause cancer (i.e., has become an oncogene) is said to be activated.

• Receptor tyrosine kinase proto-oncogenes (RTK's)

These are signal-transducers, proteins that are fixed firmly across membranes. They are present in all eukaryotes. You met them in "Biochemistry", and learned that they are activated by the ligand joining the outer surfaces of two adjacent receptors. A variety of effects (mostly turn-ons) result within the cell itself. (You can read on your own about these, or review what you have already learned about inositol triphosphate, diacylglycerol, signalling ras, etc., etc.) The stimulated receptor must be engulfed and destroyed ("receptor mediated endocytosis").

The first tyrosine kinase to be studied extensively was src, well-known in its oncogene form from a transforming virus. It has not proved much of a player in human tumors. (* We still don't know what it does for us in health.)

c-abl is translocated from chromosome 9 to the breakpoint cluster region of chromosome 22 in most cases of chronic myelogenous leukemia, and this is part of the "Philadelphia chromosome" phenomenon. Much more about this later!

fms, which codes for macrophage colony-stimulating factor receptor, is mutated in some hematopoietic neoplasms (update Br. J. Haem. 123: 749, 2003), and so is its relative fms-like tyrosine kinase 3 (FLT3).

The epidermal growth factor receptor family (EGFR/HER1, neu/HER2/erb2, erbB-2, erbB-3) (erbB discovered Science 249: 1552, 1990) are now very prominent because of the new biotechnology treatments.

As with other tyrosine kinase oncogenes, erbB mutants are those that are locked in the "on" position ("constitutively activated"). erbB-related cancers are mostly squamous cell carcinomas, and fms-related cancers are mostly hematopoietic cancers.

The related neu (once erb2, now HER2) is amplified in many carcinomas, notably adenocarcinomas, especially of the breast. Today, the degree of amplification is a guide to both prognosis and therapy.

met (HGF-receptor) tyrosine kinase proto-oncogene seems to be what produces lumens in mesenchyme and its tumors (synoviosarcomas, mesotheliomas, kidney tubules, liver tubules, others): Science 257: 1258, 1992; Cancer 82: 1513, 1998; Proc. Nat. Acad. Sci. 95: 14417, 1998; it codes for the hepatocyte growth factor receptor and is emerging as a major player (J. Urol. 170: 2163, 2003).

* Unlike most other activated proto-oncogenes, mutated met can be passed parent-to-child (hereditary papillary kidney cancer) and also arks the aggressive "tall-cell" variant of thyroid cancer.

RET is a proto-oncogene tyrosine kinase (Nature 363: 458, 1993; NEJM 335: 943, 1996), and was the first activated oncogene that was discovered being passed from parent to child (Science 267: 381, 1995). Depending on the allele, there may be various endocrine tumors, mucosal neuromas, and/or Hirschsprung's disease of the colon.

kit, the receptor for mast-cell growth factor, is yet another tyrosine kinase activated particularly in GI stromal tumors and some hematopoietic tumors (big review J. Allerg. 114: 13, 2004). As with RET and met, there are germline mutations that cause familial disease (Gastroent. 129: 1042, 2005).

* Flk-1, a VEGF receptor (activated in glioblastoma, etc., must be why they elaborate those odd blood vessels; Nature 367: 525, 1994; J. Path. 207: 224, 2005).

* PDGFRalpha is one of the receptors for platelet-derived growth factor. Mutations are common in medulloblastoma (Eur. J. Cancer 42: 646, 2006); a familial syndrome features GI stromal tumors (Lancet 369: 1731, 2007).

• GTP-binding protein proto-oncogenes

This includes the ras family, present in all eukaryotes. They code for the signal-transducing G-proteins that modulate various transmembrane signals (*  for example, turning fibroblasts into fat cells; Science 253: 565, 1991). Each codes for a p21 protein that binds GTP, and the healthy ones hydrolyze it (i.e., they are GTP-ases). They seem to be involved in initiation of mitosis as well as in differentiation. What ras does: Science 264: 1413, 1994. All about the G-proteins: NEJM 332: 406, 1995 (* Nobel prize 1994 Gilman and Rodbell).

Most oncogenic ras are mutations with a single base pair change that alters an amino acid at position 12, 13, or 61 in the protein product. This destroys GTP-ase activity but retains GTP-binding activity, and current thinking is that these stay locked "on", telling the transformed cell, "Keep dividing!"

Several of the best-known chemical carcinogens produce a specific mutation specifically at one of the three hot spots. For example, aflatoxin regularly mutates * codon 12 in K-ras (GGT to AGT or GAT).

ras oncogenes clearly help cause a large percentage of human cancers. The large majority of oncogenes isolated from human tumors have been hot-spot ras mutants. Almost all pancreatic cancers, and many other cancers (especially adenocarcinomas) have mutations in codon 12 of K-ras (update Arch. Path. Lab. Med. 126: 1096, 2002). ras activation precedes malignant expression: Science 248: 1101, 1990.

* A favorite pathology research subject in the late 1980's was immunostaining for ras p21 and seeing its effect on diagnosis and prognosis. For example, staining for K-ras p21 was reported to differentiate malignant from benign prostate epithelium; unfortunately, this has not held up on closer examination.

* We can hope for better success with new attempts to screen patients for colon polyps/cancer by checking stools for oncogenic ras; despite much discussion over the past decade, it's still too costly (Gastroenterology 119: 1219, 2000).

* gsp is the activated form of Gs(alpha), a ras-related gene that occasionally shows up altered in endocrine adenomas. More about this when we talk about the pseudohypoparathyroidism family of illnesses.

• DNA-binding protein proto-oncogenes

This is the myc family, present in all eukaryotes, whose protein products are intranuclear and bind to DNA itself. They enable DNA synthesis.

myc activation is usually by amplification (excess copies of a gene) and/or translocation rather than by mutation.

In Burkitt's lymphoma of B-cells, c-myc (chromosome 8) is moved next to the immunoglobulin gene (chromosome 14), i.e., the cell decides to multiply like crazy every time it is told to make antibodies.

myc genes are much amplified in neuroblastomas and oat cell lung carcinomas.

myb is a related proto-oncogene involved both in human cancer and in the proliferation of cells in human atheromas. * Basic biology buffs: myb has to do with production of the anthocyanins that account for the pretty colors of Indian corn.

* The Ewing's sarcoma 11:22 translocation joins FLI1, a myc-like proto-oncogene, to EWS, producing an aberrant transcription factor EWS-FTI1 (Proc. Nat. Acad. Sci. 90: 5752, 1993); much since.

• Growth factor proteins themselves

The prototype Is c-sis, which codes for the beta chain of platelet-derived growth factor (PDGF), the stuff that tells fibroblasts to divide in wound healing.

Probably sis-induced cancers grow by autocrine self-stimulation by PDGF. Not surprisingly, PDGF is greatly over-expressed in many sarcomas, and only transforms cells with the PDGF receptor.

* int-2, the second site where the mouse mammary tumor virus integrates, is the gene for fibroblast growth factor #3 (FGF3; J. Path. 170: 219, 1993; J. Neurosurg. 76: 792, 1992), a gene with many relatives, including proto-oncogenes flg (FGF1) and bck (FGF2).

• More oncogenes that are worth knowing.

raf, a family of serine-threonine kinases, is part of the signaling pathway that links most of the tyrosine kinases to the ras family. BRAF is mutated in around 15% of human cancers (Am. J. Clin. Path. 123: 256, 2005).

bcl-2, activated in most B-cell lymphomas, and its relative bcl-X, tell the cell not to undergo apoptosis, but to divide if told to do so. The molecular biology of this important molecule, and its family including bax, is being worked out. (Nat. Med. 3: 614, 1997; Nat. Med. 274: 2002).

* erbA codes for the human thyroid hormone receptor. It is linked to a variety of animal cancers (Onc. Rep. 9: 863, 2002).

* jun is the factor that initiates transcription of DNA at a particular sequence. Present in all eukaryotes. fos apparently turns short-term stimulation into long-term differentiation, and fos mutants help immortalize cell cultures. fos and jun: Science 254: 1210, 1991. Both are known from the transforming retroviruses (v-fos and v-jun). It was expected that these would turn up mutated in human tumors, but so far there's been nothing.

Cyclin D1 itself (11q13, bcl-1, the PRAD-1 locus) is involved in the oldest known lymphoma translocation, in most parathyroid adenomas, and is amplified in around 20% of breast cancers, and the knockout mice get breast tumors (Nature 369: 669, 1994). Review Oncogene 25: 1620, 2006.

The high mobility group (HMG/HMGI) genes are often scrambled specifically in benign tumors with near-zero malignant potential (Am. J. Clin. Path. 109: 251, 1998; Am. J. Path. 155: 1535, 1999). This suggests that these mutations are the single major step to producing benign tumors, which do not turn malignant because there aren't a lot of mutations accumulated.

DNA In-Situ Hybridization is beginning to come into its own as an adjunct for cancer diagnosis (Am. J. Clin. Path. 112(S1): S11, 1999).

* Cytogenetic analysis of tumors by making karyotypes was never much use, since until recently it was impossible to keep the stromal cells and lymphocytes from overgrowing the cancer cells in culture. (Of course, this is also a reminder that cancer cells aren't dividing that rapidly; they're just dividing without the usual controls.)

Today's techniques count chromosomes instead by fluorescent means, which stain each pair a different color.

It is now standard to stain paraffin sections of breast cancers with probes for erb-B2 (HER2/neu). Cancers without amplification will show only two loci; those with amplification will show huge numbers of loci, usually in big clumps.



When there's suspicion that a particular gene is split by translocation, two probes can be used, of different colors. If they remain together, there is no translocation; if they are separated, translocation has occurred. This is very helpful in diagnosing Ewing's sarcoma.

TUMOR SUPPRESSOR GENES (anti-oncogenes): Many reviews; Knudson himself in Proc. Nat. Acad. Sci. 90: 10914, 1993; at the bedside Lancet 349-S2: 16, 1997; kids Ped. Clin. N.A. 49: 1393, 2002; adults Arch. Path. Lab. Med. 125: 85, 2001.

Tumor-suppressor genes keep cells from overgrowing, even when the oncogenes are activated. To lose their anti-cancer effect, both copies must be altered. (Contrast the proto-oncogenes, which exert their effect when a single copy is activated to an oncogene.)

Knudson's Law for tumor-suppressor genes

One hit: You have a cell with a much increased propensity to turn neoplastic

Two hits: You have a tumor cell.

If you inherited one copy of the damaged tumor suppressor gene, you have the anti-oncogene deletion syndrome, with a greatly increased risk for the corresponding tumor(s). If you have the corresponding tumor(s) but do not have the germ-line mutation (i.e., your tumor was sporadic), both mutations are somatic. If you don't understand this, stop now and think about it until you do.

Please don't ask whether the mutated allele is "dominant" or "recessive". It is dominant with respect to the tumor-family syndrome, recessive with respect to the tumor itself. If you are reading this, you already understand.

If you derive from a mutation-bearing sperm or egg, or were hit at conception, you have one of the autosomal dominant tumor suppressor gene deletion ("tumor-susceptibility") syndromes. The malignant phenotype requires both copies to be bad, so it is autosomal recessive.

* Tumor suppressor genes are often "mutated" by loss of methylation of their promoters: "cancer epigenetics". Nat. Genet. 21: 163, 1999; Nat. Rev. Genet. 3: 415, 2002.

The prototype Is retinoblastoma, a hereditary cancer.

The tendency to retinoblastoma (*  and osteosarcomas, in survivors) is inherited as an autosomal dominant trait, i.e., there is one chromosome lacking a particular tumor suppressor gene.

Retinoblastoma results when a mutation damages both copies the healthy tumor-suppressor gene in a single cell (the Knudson "two-hit" phenomenon). The cell now lacks any copy of the tumor suppressor gene, and is transformed.

The retinoblastoma susceptibility locus is RB1 at 13q14. The protein is central to cell cycling. Update on how it works: Nature 374: 114, 1995.

The most important finding in cancer research in the late 1980's was the discovery that many (if not most) DNA-containing tumor viruses (including human papilloma virus) bind to, and inactivate, the normal product of RB1 (Cell 56: 1, 1989). The ability to do this correlates with tumorigenicity (Science 248: 70, 1990); virus protein blocks the binding of Rb to its target among the nucleoproteins (Nature 351: 406, 1991), etc. This has held up, though to date it has not resulted in any new therapeutic strategies.

The most common known genetic injury in human cancer is damage to the p53 (TP53) gene.

*  "Molecule of the year" cover story Science Dec. 24, 1993; update Nature 370: 174, 1994, review Lancet 346: 1009, 1995, Br. J. Surg. 85: 1460, 1998; stains for the tumor product are now part of cancer prognosticating: Am. J. Clin. Path. 113(5S1): S84, 2001; Arch. Path. Lab. Med. 124: 966, 2000; lots more).

This is the "most-often-mutated" known gene in spontaneous human cancers (Nature 350: 377, 1991), a dubious honor once held by ras. Older reviews of p53: Nature 350: 429, 1991; Nature 351: 453, 1991; Science 253: 49, 1991. Mutations vary; unlike ras, the only "trademark" is the hepatocellular carcinoma mutation due to aflatoxin (codon 249).

The p53 gene product is a sequence-specific binder to DNA that prevents mitosis during times of cell injury, so that there will be more time for DNA repair. Sometimes (gamete, lymphocyte, chemotherapy) p53 even tells an injured cell to undergo apoptosis.

* The protein binds as a tetramers, making it easier to understand the oncogene-anti-oncogene duality. The defective unit handcuffs the others. See Science 256: 827, 1992; Nature 358: 15, 1992.

A famous link is to hepatocellular carcinomas (*  those from the aflatoxin-hepatitis B belt have a distinctive mutation in codon 249: Science 253: 49, 1991; the mutagen is aflatoxin: Lancet 338: 1356, 1991).

More good evidence for the Nowell multi-step clonal evolution model for tumorigenesis is provided by the fact that astrocytomas (low-grade brain cancer) turn into glioblastomas (high-grade brain cancer) upon acquiring a mutated p53 (Nature 355: 846, 1992).

* UV light mutates p53 in skin cancer: Proc. Nat. Acad. Sci. 88: 10124, 1991). "Good" p53 might reverse myeloid leukemia in individual cells (Nature 352: 345, 1991), and so forth.

The tumor-suppressor gene p53 product is also inhibited by the proteins of human papilloma virus (see above; p53 and cervix cancer Lancet 340: 140, 1992), SV40, and adenovirus 5 (Science 248: 76, 1990).

We now stain for the expression of this protein (the damaged genes make more of it) in cancers for prognostication (Arch. Path. Lab. Med. 121: 395, 1997.)

There are many other tumor-suppressor genes.

As noted above, p16INK4a (was CDNK2A or MTS1), on 9p21, is an inhibitor of cyclin-dependent kinase (i.e., inhibitor of mitosis) which is very commonly deleted in lots of cancers (Proc. Nat. Acad. Sci. 91: 11045, 1994); it may function either an oncogene or a tumor suppressor gene. See also Nature 370: 180, 1994; Nat. Med. 5: 731, 1999; lots more. This now identifies a familial melanoma syndrome: NEJM 338: 879, 1998. All about cell-cycle screwups in cancer: Science 266: 1821, 1994.

Apparently all renal cell carcinomas have lost the Von Hippel-Landau locus (Nature 332: 268, 1988; more below). Oat cell carcinomas also lack a portion of 3p; almost all other lung cancers lack a smaller chunk in the same place.

People heterozygous for a deletion of a bit of 11p have aniridia, and they develop Wilms' tumors that are homozygous for the deletion (as are spontaneous Wilms' tumors: Proc. Nat. Acad. Sci. 90: 1416, 1993).

This is the famous WT-1 (formerly WAGR) locus, producing a nuclear binding protein that seems to be involved in mesenchymal-epithelial transitions in general, and kidney and urogenital ridge development in particular (Nature 346: 194, 1990; Nature 353: 431, 1991; Proc. Nat. Acad. Sci. 90: 5828, 1993). * Biology jocks: the adjacent aniridia gene is homologous to the drosophila homeobox PAX6.

Von Recklinghausen's disease type I patients are heterozygous for a mutation on chromosome 17; the locus is "NF-1", and its protein product, "neurofibromin" characterized as a facilitator of hydrolysis of GTP by normal ras p21 (Nature 351: 576, 1991).

Von Recklinghausen's disease type II patients are heterozygous for a mutation on chromosome 22 (product is * schwannomin or merlin), and their tumors (and many spontaneous meningiomas and other nerve tumors: Am. J. Hum. Genet. 47: 823, 1990) are homozygous for this loss. See below.

Colon cancers typically become homozygous for deletions of 5p (early) and 5q21 ("APC", "adenomatous polyposis coli", "familial polyposis locus".

In any case, the study of colon cancer progression has provided strong support for the Nowell multi-step clonal evolution model of carcinogenesis (no surprise). The colon is a handy organ since pre-cancerous polyps often make it to the lab.

* A gene on chromosome 2 makes a person susceptible to melanomas of the eye.

The familial non-polyposis ("Lynch") family of tumor-suppressor genes are responsible for repair of DNA mismatches. * Alphabet soup: hMSH2, hMLH1, hPMS1 hPMS2, and so forth; MLH1 can be inherited as a mutant with normal sequence but abnormal methylation NEJM 356: 697, 2007.

We now screen for the entire family using microsatellite instability assays. See below.

* A new genetic mechanism in the development of cancer is probably relaxation of imprinting. Wilms tumor is the only cancer I've heard about so far (Nature 362: 747 & 749, 1993; Cancer Res. 60: 2356, 2000).

* So far, the only tumors in which changes in mitochondrial DNA have been detected are "oncocytomas", i.e., tumors made up of cells with preposterous numbers of mitochondria. More about this under "Kidney". See also Am. J. Hum. Genet. 52: 537, 1993 (mitochondria and lipomas).

During tumor progression, the second anti-oncogene is lost by nondisjunction and loss of heterozygosity when an extra chromosome is randomly eliminated. The phenomenon enabled the discovery of many of these genes.



* Actually, we've known for decades that tumor suppressor genes must exist. A classic finding is that when a cancer cell and a non-cancer cell are fused, the product is often a non-cancer cell.

LATE PROGRESSION

The subject in basic cancer research right now.

Once its growth genes have been mutated and its genome perhaps destabilized, cancer is still not a threat until it has developed the ability to invade, to induce its own blood supply, and (usually) to spread to distant sites.

We are just now unravelling the additional accumulated mutations that allow these things, and they will be the targets of new cancer therapies.

A major mystery of medicine is why some cancers (notably metastatic breast carcinoma and metastatic melanoma) lie dormant and hidden for years (even decades).

For that matter, it's puzzling why cancer metastases will thrive in one locale and not in another.

* Virchow actually doubted that cells travelling from the cancer are responsible for metastases, since lung metastases were less common than liver metastases in breast cancer.

* Your genetic makeup determines whether cancer will progress more or less rapidly (Am. J. Hum. Genet. 63: 1, 1998), etc., etc. The genes remain elusive, but this promises to be interesting.

Microsatellite instability means the presence of abnormally long stretches of DNA with repeating motifs of 1-5 base pairs (i.e., a long string of "A"'s or "AC"'s or whatever), indicating defective DNA repair. This can actually be present quite early, before there are any visible histologic changes.

Today, pathologists routinely get microsatellite instability assays on certain cancers (colon, endometrium, ovary; especially certain histologic types and those occurring in the young.) A tissue stain for the mismatch repair proteins seems to be almost as good as the actual genetic assay: Cancer 95: 2422, 2002; Arch. Path. Lab. Med. 127: 694, 2003. The purpose is to screen for Lynch's syndrome / Lynch-like new mutations, and (at least for colon) as a guide for prognosis (better than the rest) and treatment: NEJM 349: 247, 2003. Lynch cells are "RER-positive" (replication error positive).

Chromosomal instability of course means there have been problems with the mitotic spindles and non-disjunction will become commonplace. Cells without the necessary chromosomes die and don't contribute to progression; cells with extra copies may possess a growth advantage. Eventually this leads to those variably-sized and often huge nuclei that are familiar from conventional histology of cancers (Am. J. Path. 161: 391, 2002; many others.

TUMOR IMMUNOLOGY

Despite a great deal of data and the fact that lymphocytes are usually abundant in advancing cancers, study of the immune system's attack on tumors has yielded little useful information.

For some cancers (breast cancer, maybe melanoma, etc., etc.), a dense infiltration of lymphocytes imparts a slightly better prognosis, but this is hardly decisive. (The most impressive link so far is for ovarian epithelial carcinoma: NEJM 348: 203, 2003).

By and large, cancer treatments that focus on strengthening the immune system (BCG, cytokines) have failed to help most cancer patients.

* In the early 1990's, there was considerable interest in cultivating and re-infusing lymphocytes found in tumors (i.e., lymphocytes that are presumably attacking the tumor). This "autologous mature lymphocyte transfer" sometimes resulted in some regression of the tumor (Cancer Res. 51(S): 5074, 1991). The main problem was that the re-infused lymphocytes tended to just die. It's still done occasionally, especially for melanoma (J. Immuno. 173: 7125 & 7622, 2004).

Despite the widespread belief (medical, folk) that "our immune system is our main defense against cancer", evidence in support of the "immune surveillance theory" is not very convincing.

People who are immune-crippled (hereditary immunodeficiency, AIDS, immunosuppressive therapy) simply don't have increased rates for any of the common cancers. See, for example, Chest 103: 862, 1993; review in AIDS: JAMA 285: 3090, 2001; JNCI 28: 5-9, 2001; update Lancet 370: 59, 2007 (the cancers that aren't caused by infectious agents just aren't increased in AIDS or iatrogenic immunosuppression).

The malignant lymphomas that are result from immunosuppression (AIDS, transplant patients) arise from cells that are already undergoing compensatory hyperplasia because of the immune suppression. Actually many of these are not even tumors, but hyperplasias that regress when good immune function is restored.

Kaposi's "sarcoma", seen in AIDS and transplant patients, is actually a viral hyperplasia, not a real tumor.

This has been obvious for decades. It arises multifocally, lacks anaplasia, is common among the immunosuppressed, and is epidemic in parts of the world.

Patients with ataxia-telangiectasia (cancers plus immunodeficiency) have "fragile chromosomes" as the basic problem; even carriers are at extra cancer risk.

Carcinomas of the skin and lower lip are common in transplant patients and they tend to be aggressive. This is not true of other carcinomas, and rather than invoking "escape from immune surveillance", the usual suspects are known and unknown viruses (Cancer 85: 1758, 1999).

Nude mice have no cell-mediated immunity but are no more susceptible to spontaneous cancers than any other mice.

People with autoimmune diseases triggered by the known anti-tumor antibodies do not self-cure their tumors (though there may be modest regressions), nor do these diseases pop up in the absence of the tumors known to cause them (Lancet 341: 21, 1993).

Frankly, tumor immunology is only starting to get interesting to patient-care physicians. Although "cancer is not other, it is us", a few known antigens are distinctive (though not unique) for individual cancers, and it is against these that any effective immunotherapy will probably be directed.

* Today, people talk about "immune escape", which happens quickly once the genome is destabilized: Nat. Immuno. 3: 999, 2002 (acknowledge that the classic "immune surveillance" hypothesis is completely discarded); Nat. Immuno. 3: 991, 2002. Immune surveillance updates: Nat. Rev. Cancer 2: 850, 2002.

* One very well-known phenomenon is that in allogenic bone marrow transplantation to cure cancer, a strong graft-vs.-host response increases the likelihood of an actual cure. It seems most likely that the donor cells are attacking the last of the cancer cells because they are slightly different from the rest of the host's minor incompatibility antigens, to which they have tolerized. Update on hematopoietic stem cell transplantation: NEJM 354: 1813, 2006.

* One interesting approach tags antibodies against such antigens with pseudomonas exotoxin ("immunotoxins"), selectively destroying cancer cells. Minimal side effects. So far it's proved helpful for a few human cancers (notably hairy cell leukemia): Blood 94: 3340, 1999; Nat. Biotech. 16: 449, 1998; however, progress is held up by the tendency to make antibodies against the agent (J. Imm. 177: 8822, 2006).

* Another interesting approach will be injecting the patient with his or her own cancer cells crippled by an anti-sense gene against the tumor's autocrine growth factor (i.e., they are immunogenic but do not grow well: Ann. Surg. Onc. 8: 32, 2001, others). These cells are immunogenic but are much less tumorigenic, and there's a claim of complete remission of gliosarcomas in mice (Proc. Nat. Acad. Sci. 93: 2909, 1996).

Melanoma has a couple of antigens that are attacked by the immune system ("MART" -- "melanoma antigen recognized by T-cells"; also HMB-45/gp100 which is a popular stain to diagnose melanoma, Cancer 93: 409, 2001). Neither expressed much in health, but comes out in activated and malignant melanocytes. There are some others too (J. Imm. 168: 951, 2002). Melanoma and renal cell carcinoma are unusual in that generalized immune stimulation (i.e., by interleukin 2) actually helps sometimes (Cancer 91: 806, 2001).

Despite the near-total failure of tumor immunology itself to help cancer patients, the era of magic bullets (targeted therapies, i.e., specific monoclonal antibodies and specific oncogene product inhibitors) in the treatment of cancer has arrived. Today's arsenal includes:

• rituximab ("Rituxan", anti CD20 monoclonal antibody)
• cetuximab ("Erbitux"; anti-HER-2/neu monoclonal antibody)
• trastuzumab ("Herceptin"; anti-HER-2/neu monoclonal antibody)
• pertuzumab ("Omnitarg"; anti-HER-2/neu monoclonal ant