Cancer and Your Genes

Mechanism of Cisplatin-Resistance in BRCA2-Related Ovarian Cancers

Matt Mealiffe, M.D. — Sun, 10 Feb 2008 17:53:03 -0800

Our increasing molecular knowledge of the pathways involved in cancer development continues to suggest potential avenues for molecularly-targeted cancer therapies. However, aside from a few examples (Gleevec, Herceptin, and others), molecularly-targeted therapy is still not a reality for most cancer types.

Individuals born with mutations in the BRCA2 gene have a substantially elevated lifetime risk for breast cancer, ovarian cancer, and several other cancers. Recent work has shown that loss of BRCA2 function results in defects in a DNA repair process called "homologous recombination." Evidence has emerged that PARP-inhibitor drugs, which are currently under development as possible treatments for BRCA2-associated cancers, are potent killers in vitro of cells lacking BRCA2. In fact, the loss of homologous recombination capacity in BRCA2 mutant cells is thought to be an achilles heel that may be exploited both by PARP-inhibitors and also by platinum compounds such as cisplatin (BRCA2-mutated ovarian cancers seem to be particularly sensitive to cisplatin).

However, individuals treated with another molecularly-targeted therapy, Gleevec, for chronic myelogenous leukemia have developed resistance to the drug in some cases. Indeed, individuals with BRCA2-related ovarian cancers ultimately develop cisplatin resistance.

Two papers published online today in the journal Nature (abstracts available here and here) show us why this is the case at least in a subset of cases in individuals with BRCA2 mutations. Basically, various revertent mutations can occur under selection of cisplatin or PARP-inhibitors that result in expression of a functional BRCA2 protein again that improves the cellular capacity to promote homologous recombination, thus eliminating the achilles heel. Thus, BRCA2 loss is involved in the establishment of the cancer, but is not necessary for its maintenance (especially under the selective pressure of treatment with Cisplatin).

This has a number of implications:

Genotyping of tumors in individuals with BRCA2 mutations may help to predict whether PARP-inhibitors or platinum-containing compounds are likely to be effective in their treatment
BRCA2-associated tumors that have acquired PARP/cisplatin-resistance may be treatable with a combination of cisplatin or a PARP-inhibitor plus a secondary drug disrupting homologous recombination
Although PARP-inhibitors have not yet reached the clinic, we now know of one potential challenge in their potential use
As some sporadic ovarian cancers seem to have lost BRCA2 function, these findings may be relevant to understanding treatment resistance even in some ovarian cancers that apparently are not hereditary

Childhood Cancer and Birth Defects

Matt Mealiffe, M.D. — Sat, 05 Jan 2008 18:52:44 -0800

It is now well known that certain clinical genetic syndromes recognized in childhood are associated with an increased risk of cancer development. For example, in the Gorlin syndrome (also known as Nevoid Basal Cell Carcinoma Syndrome), which may affect ~1 in 40,000 individuals, a mutation in the PTCH gene leads to a number of unusual features including multiple jaw cysts, a very large head circumference, and prominent foreheads in addition to skeletal (bifid ribs and wedge-shaped vertebrae) and other anomalies. Although the causative mutation in PTCH leads to these morphological abnormalities, it also leads to a substantial increase in cancer risk, primarily for skin basal cell carcinomas and medulloblastoma.

Although a number of studies have tried more broadly to assess the association between various birth defects (major malformations and minor anomalies) and cancer, these have been plagued by limitations including the reliance on chart reports of the morphological abnormalities.

A new study, reported in the January 2 issue of JAMA, provides an intriguing new look at this issue.

The authors personally examined a total of 1073 childhood cancer patients and 1007 controls individuals (all patients and controls were of northern European descent) without cancer with an eye toward cataloging "morphological abnormalities" (both major birth defects and minor anomalies) in a standardized fashion.

Interestingly, they found that both major birth defects and minor anomalies were significantly more common in the pediatric cancer group as compared to controls. For example, per 1000 individuals, 268 major abnormalities were found in cancer patients vs. 155 in controls; likewise, there were 1252 minor anomalies per 1000 individuals in cancer patients vs. 898 per 1000 controls. This was statistically significant and is consistent with the notion that at least a subset of pediatric cancers are associated with genetic defects that might also predispose to combinations of birth defects and/or minor congenital anomalies.

In 42 patients, there was an already established clinical genetic syndrome. Even after these patients were removed from the analysis, there was still a significantly higher number of major abnormalities and minor anomalies in the cancer patients as compared to controls.

After excluding the 42 patients with preexisting diagnoses, the authors sought to assess which congenital anomalies seem to be particularly associated with pediatric cancer. For example, they showed that blepharophimosis (a static reduction in the distance between the upper/lower eyelid resulting in a narrowed slit-like appearance), a minor anomaly, was about 11-times more likely to be found in pediatric cancer patients as compared to controls. Some of the other abnormalities that were statistically associated with pediatric cancer were asymmetric lower limbs and broad feet (appearing disproportionately wide for length and for which the measured width is >95th percentile for age).

The authors then went on to make some initial attempts to identify patterns of morphological abnormalities to classify them into putative new syndromes.

Overall, this study is very intriguing, but has some methodological limitations that will require validation of the result in an independent group of patients and controls. Nevertheless, this is a very plausible result supported by the existence of a number of known genetic syndromes in which certain birth defects cluster with certain types of cancer.

This study provides a foundation on which future efforts to identify new tumor predisposition syndromes will be based. Ultimately, this should allow improved efforts to screen for and/or prevent some childhood cancers.

Colon Cancer Risk: More Evidence for Common Disease, Common Gene Hypothesis

Matt Mealiffe, M.D. — Tue, 20 Nov 2007 21:36:07 -0800

We know from both population-based epidemiologic studies and from twin studies that our genes make a substantial contribution to risk for colorectal (aka colon) cancer. Although there are several relatively uncommon familial colon cancer predisposition syndromes (which account for <5% of all colon cancer) with known genes involved, progress towards unraveling genomic variants involved in the more common, run-of-the-mill colon cancer cases has been slower in coming.

However, a study published recently in the journal Nature Genetics reports significant progress in this regard. This study is one of a flurry of genome-wide association studies (GWAS) - which utilize a study design made possible by recent advances in our ability to simultaneously examine an individual's genotype at hundreds of thousands of sites of single nucleotide polymorphisms (SNPs).

Pursuing the idea that variation in colon cancer risk may be due to relatively common low risk SNPs (perhaps in combination), many of the same authors (a group led by Richard Houlston and Ian Tomlinson) had reported in the August 2007 Nature Genetics that a SNP located on chromosome 8 was associated with colon cancer. Although statistical support was strong, it is important to keep in mind that the risk to individuals with the variant SNP (whether they have one copy or two) was modest. Odds ratios were 1.27 for heterozygotes (with one copy of the risk variant) and 1.47 for homozygotes (who carry the risk variant on both copies of chromosome 8). There were also reports from other groups in Nature Genetics about colon cancer risk and this region of chromosome 8.

Now, in the new paper published online in the November 2007 issue of Nature Genetics, the group led by Houlston and Tomlinson focused on another strong signal from their genome-wide association study. They identified three SNPs in a gene called SMAD7 that were very significantly associated with colon cancer risk. Interestingly, as with the results from a number of other recent GWAS studies, the significant SNPs were intronic (i.e., located in an intron, in between the exons which are the stretches of DNA with the information coding for amino acids, the protein building blocks). As is often the case, the functional SNP at this locus is unknown at this point.

SMAD7 is also known as "mothers against decapentaplegic homolog 7." Unfortunately for those of us who have to deal with this stuff clinically, the drosophila (fruitfly) genetics community has a tradition of picking names of this sort. Nevertheless, the available information about SMAD7 function within the cell suggests that it is quite plausible as a colon cancer risk gene. Specifically, it acts as an antagonist of TGF-beta signalling, a pathway known to be relevant to colon cancer development.

The authors suggest that the SMAD7 risk variants likely contribute to about 15% of colorectal cancer cases. For those of us practicing clinical cancer genetics, however, it is important to note that SMAD7 seems to be involved in less than 1% of familial colon cancer though. This fact, in conjunction with the relatively modest odds ratios, will likely limit applicability of this new result to the individual patient. Nevertheless, I can certainly see a future coming in which the SMAD7 SNPs - presumably in conjunction with other colon cancer risk SNPs - are utilized on a population-wide basis to partition individuals into different risk categories with different recommended screening strategies.

Medicare and Genetic Testing for Cancer Risk

Matt Mealiffe, M.D. — Sat, 20 Oct 2007 13:45:57 -0700

I just received a letter from Myriad Genetics regarding likely changes in the Medicare policy for reimbursement of genetic testing. The letter summarizes a number of proposed changes to this policy which are likely to go into effect beginning Nov. 1.

There's some good news here:

Inclusion of criteria for testing of the MSH6 gene in Lynch syndrome (aka Hereditary Non-polyposis Colorectal Cancer [HNPCC] syndrome)
In order for genetic testing to be reimbursed, documentation requirements will include an Informed Consent document that indicates that the patient has been informed of a number of specific issues and agrees to post-test counseling

There is some news that is perhaps not so good:

There will be a clear-cut requirement that there be documentation in the medical record and/or office notes that the genetic testing is intended for the medical management of the patient (i.e., the patient sitting in front of the physician or genetic counselor, not their family members). This is potentially bad news for those who feel strongly that genetic information has substantial value beyond that which immediately affects clinical care and decision-making. However, it is not surprising given the enormous cost pressures on Medicare and the ways in which we fund and evaluate medical care currently in the U.S.
Testing of unaffected (i.e., no personal history of cancer) individuals will not be covered. This is potentially a big problem. And many will argue that it is a misguided rule change. Although the Medicare population is 65 and older and perhaps less likely to benefit from preventive options guided by genetic testing, this proposed change will negatively impact care of families guided by genetic testing in a very important way:

A common scenario in the cancer genetics clinic might involve a woman who is in her mid-30s and concerned about her risk for breast and ovarian cancer, because one of her paternal aunts died of premenopausal breast cancer and another of her father's sisters died of ovarian cancer diagnosed in her 40s.

A cardinal rule in clinical cancer genetics is that you always want to do the genetic test on someone in the family who has had a cancer diagnosis that is suspicious for the risk syndrome that you are considering. If either of this woman's aunts were alive, they would clearly be the best people to test first.

However, they both died of their disease. In the absence of other individuals with cancer to test, the patient's healthy father would be the next best person to test. The reasoning is as follows:

If we just test that woman in clinic for BRCA1 and BRCA2 mutations and the test is normal, we don't know whether this is because: 1) other family members have a detectable BRCA1 or BRCA2 mutation that she did not inherit (in this case, her cancer risk in the above scenario would be no different than that of any other woman) or 2) the early breast and ovarian cancer cases in the aunts might have occurred in the absence of any known BRCA1 and BRCA2 mutations (in this case, the test for the patient in clinic is not informative).

If the patient's aunts had a BRCA1 or BRCA2 mutation, then there is a 50 percent chance that her father will have it as well. Therefore, by testing him first, it can make the information gained by the patient's BRCA1 and BRCA2 testing more helpful.

As we move toward a future of more predictive and preventive medicine incorporating genetic/genomic information, there will continue to be tension at the intersection of the prevailing medical/reimbursement model that focuses on the patient and attempts to incorporate genetic information which is most often more valuable in the context of the whole family. Ironically, to realize the full potential of personalized medicine, it will be essential to resolve these impediments to care of the whole family.

Genetic Voyeurism on TV at "DNA and You"

Matt Mealiffe, M.D. — Sat, 20 Oct 2007 13:40:44 -0700

This post about a particularly public case of BRCA1/BRCA2 testing may interest readers of Cancer and Your Genes.

Cancer and Your Genes

Matt Mealiffe, M.D. — Sun, 14 Oct 2007 15:14:46 -0700

Welcome to my new blog, "Cancer and Your Genes." I'm also the author of "DNA and You" - a personalized genomics focused blog located here.

As a clinical cancer geneticist (with board certification in both clinical genetics and internal medicine), I have become more interested in creative means to educate the public about the contribution that our genes make to cancer risk. I intend for this to be a dialogue and aspire to present solid, well-reasoned, and, hopefully, interesting discussion of important emerging issues in this area.

It's important to point out that the information presented here does not substitute for care from your personal physician. Likewise, reading this blog does not constitute the establishment of a physician-patient relationship between us. If you would like more information, I would encourage you to schedule an appointment to see a clinical geneticist or genetic counselor in your area. A good place to find a genetic services provider in your area is the Clinic Directory at the GeneTests website.

In any event, I hope that you both enjoy this blog and find the information useful.