False positive paradox and results from genetic testing

A friend recently had a genetic test done that showed she has a BRCA mutation, the same diagnosis that persuaded Angelina Jolie to have a mastectomy. She’s been told that the result indicates that she has a 40% greater chance of developing breast or ovarian cancers.

While surgery seems like a reasonable solution, I’m curious about how the false positive paradox might affect her decision, and I’m having a difficult time finding reliable figures to make the calculations. Any help or insight would be greatly appreciated.

If your friend is concerned about a false positive, she can just have the test repeated, no?

My wife died from ovarian cancer and we didn’t discover she had the BRCA mutation until after the cancer was diagnosed. I think you may have misunderstood the risk number you quote. My understanding is the risk of ovarian cancer is not 40% highter but many times higher for a woman with the BRCA mutation than for a woman without it. Among the general population, about 1.5% of women develop ovarian cancer, while among women with a BRCA mutation, about 50% develop it. Ovarian cancer is nasty because it has few and very subtle symptoms until the cancer is quite advanced. If your friend does not have preventive surgery, she should be very alert to any extended periods of what appears to be an upset stomach.

As a general rule: That depends on what caused the false positive. Depending on the cause, it might just recur no matter how many times the test is repeated.

I don’t know whether that applies to this particular test.

She can have the test repeated. A complete genetics study should be conducted on her. I’m sure her doctors will be asking for a family medical history and evaluating every aspect of her health further before any surgery. Given enough indications of potential cancer developing she’ll have to weigh the odds and decide. Not everyone who discovers this mutation has the surgery. The false positive potential is just part of the risk calculation.

However well meaning, I think attempting to offer amateur advice in this situation is not a good idea, and potentially distracting when your friend has some critical decisions to make. My professional career was in genetics, and I wouldn’t try. The issues that you’re concerned about are the specific purview of a professional genetic counselor. If she has taken the test, she should already have one. If she’s finding it hard to absorb everything, I’d offer to sit in on a consultation with her, prepared with all your questions, and go from there. You can certainly grill a genetic counselor on the data supporting their advice, and wade into the latest research with their guidance if you wish. And don’t hesitate to get a second opinion if your friend doesn’t feel adequately informed.

Standard laboratory procedure - required by regulations - is to confirm any pathogenic variant detected with at least two different, independent methods before reporting it. People running these labs (myself included), understand the importance of making sure results are accurate, particularly when it’s something as major as this.

I spent the last year or so running a medical genetics lab that did BRCA testing, among other things.

It’s worth noting that the “false positive paradox” arises from a situation where the condition being tested is rare, and the accuracy of the test is not that high. Essentially, if a test’s false positive rate is similar to or higher than the incidence of the tested condition, it’ll be a lousy screening tool. It’s mostly a “paradox” because our intuition of a figure like “98% accuracy” is not very good. Instead, compare that test’s 2% false positive rate with the true incidence which might be <0.1%.

In the case of a genetic test ordered by an doctor because of a family history of a disease with well-known genetic risk factors, the patient population will have a much higher incidence of high-risk genetic variants than the general population. And the false positive rate for genetic sequencing and other genotyping methods is very low.

Smeghead, do you have any figures on the typical false positive rate, positive predictive value, etc. for a clinical genetic test? My google-fu is mostly pulling up pop science articles warning of the false discovery rate in genome-wide consumer genetic tests, which is a whole 'nother matter…

All I can say was that in our lab, for all of our validation and proficiency testing, we were at 100%. We never, to our knowledge, got a false result. Each lab is responsible for their own statistics and should have or make that information available to anyone who asks.

For validation testing, we ran something like 30-40 samples, all with known variants tested and reported by other sources. We got matching results for all of them multiple times. This is part of the validation process all labs have to do before they can start using a test clinically.

Proficiency testing involved a similar process - we received typically three anonymous samples twice a year that we’d test, then report our results to the proficiency testing company. Again, we passed with 100% accuracy during the time I was there.

This is typical of how any decently run reputable lab should be run. In fact, I’d say it’s the bare minimum. ANY result that later turns out to be wrong will prompt a thorough investigation as to what failed and how, concluding with corrective action to make sure that it never happens again. And for something as well-established as BRCA testing, a bad result should be extremely unexpected and unlikely.

I’m not saying that mistakes never happen - we’re all human and they do from time to time - but unless there’s a concrete reason to suspect the result is incorrect (such as mother and child having different pathogenic variants or a male testing as female or something), then there’s no real cause for worry. If paranoia won’t let you rest, get tested again, but personally, I wouldn’t bother.

Yeah, the more I think about failure modes for something like Sanger sequencing, the more it seems that the only way to get a false positive is by switching samples. Which happens very occasionally in a research lab, where samples are tracked with cryptic short hand, written in chickenscratch, using ink that rubs off the vial (and where I’m mocked for going to the trouble of labeling a vial on the top AND on the side…)

I’m going to hazard a guess that clinical sample handling and record keeping is rather more rigorous than that.

Hereis an FAQ from the CDC in 2003 that claims the error rate in BRCA genetic sequencing is 0. It says pretty much the only way to get a false positive is mishandling of sample.