I find myself increasingly unsettled by the manner in which some non-invasive prenatal screening (NIPS) laboratories are choosing to handle their non-reportable results. Imagine that your friend is pregnant with her second child. Her obstetrician speaks to her about NIPS, and she decides to proceed. The result comes back as non-reportable, and she is referred for genetic counseling to discuss the options of (1) doing nothing further, (2) repeating the NIPS, or (3) proceeding with prenatal diagnostic testing. In preparation for her visit, your friend’s genetic counselor calls the NIPS laboratory to follow up on the non-reportable result. The laboratory informs the genetic counselor that the data suggest a deletion on one of the chromosome 15s, but that since the test is not validated for this abnormality, it instead triggered a non-reportable result. And herein lies the conundrum. The laboratory has now disclosed information about a possible, unvalidated, incidental finding to the genetic counselor after issuing a clinical report stating that an interpretation of the data could not be given. And we wonder why the US Food & Drug Administration (FDA) has decided its time to start regulating laboratory developed tests…
The FDA and Laboratory Developed Testing
In July 2014, the US Food & Drug Administration (FDA) notified Congress of their intention to implement a risk-based regulatory framework for Laboratory Developed Tests (LDTs). The FDA defines an LDT as one that is developed, manufactured, and run within a single laboratory. LDTs vary in their level of technical complexity and in their level of associated “risk”, which is typically gauged by the severity of the anticipated or potential consequences of an erroneous result. (In 2012, the American College of Medical Genetics and Genomics (ACMG) released a policy statement to provide a specific framework for risk classification of inherited disorders.) The response to the FDA’s proposal to regulate LDTs by the medical community as a whole has been largely negative. In November 2014, the American Medical Association partnered with a number of professional societies, laboratories and laboratory directors and sent a letter to the FDA indicating that the proposed regulatory framework was in significant conflict with existing regulations, and that the FDA’s legal authority to regulate LDTs was in fact, questionable. There are certainly merits to the arguments on either side of the issue, however, my point is that the practice of reporting of incidental, analytically and clinically un-validated results as part of clinical care is just the sort of behavior that makes LDTs a legitimate target for the FDA.
Non-Reportable NIPS Results
In NIPS, a non-reportable result typically indicates that the laboratory is unable to make a determination regarding the risk of abnormalities for which the test was designed to detect, based on the data provided. There are a number of causes of non-reportable results, and patients are advised that this is a possibility prior to undergoing NIPS (hopefully). What I take issue with is the fact that a follow-up phone call from a clinician might mean being provided with some sort of quasi-result – an unvalidated, unsubstantiated result that lacks peer-reviewed evidence to support the interpretation of the data. To me, this is akin to providing a patient with results from testing that is still in the research & development stage, but treating it as though it were information from a clinically-validated test, upon which treatment decisions can be made. While several NIPS laboratories have policies regarding the ability to opt in/out of receiving additional information about certain microdeletion syndromes or trisomies commonly seen in pregnancy losses, I was unable to find any policies regarding how the discussion of non-reportable results are handled. Without clear, publically-available policies regarding what type of incidental and/or “investigational” information will or will not be reported, clinicians who decide to follow up on non-reportable NIPS results risk being put in the uncomfortable and ethically murky position of receiving an unvalidated result and having to decide whether it is most appropriate to withhold this information from the patient and simply inform her of the need to repeat the test, or to share the burden of this knowledge, hoping that the patient will fully understand the significance of a “result” that has been neither analytically nor clinically validated.
It’s Not a VOUS
For those of you who would argue that providing information about unvalidated finding is similar to encountering a variant of uncertain clinical significance (VOUS) on a prenatal microarray, I wholeheartedly disagree. A VOUS result on is one that is obtained on a platform with established analytic and clinical validity. With a VOUS result on microarray, it’s not the copy number variant call that is in question; it is the clinical significance of that copy number variant and how it may or may not impact clinical care that is in question. In contrast, if an NIPS laboratory provides information about a copy number change for which their platform has not been analytically validated, it is the result itself that is in question: is this accurate information or inaccurate information?
In the nature of full disclosure, I find NIPS fascinating in part because I work in a clinical laboratory that performs prenatal diagnostic testing. Over the past couple of years, we have handled a significant number of discordant NIPS vs. CVS/amniocentesis results, particularly with respect to suspected microdeletions or partial monosomies. It is clear that we are gaining insight into some of the biological explanations for these discrepancies (vanishing twin; fetal, placental or maternal mosaicism; maternal neoplasm, etc.), however, we in our laboratory have observed that regions of homozygosity seem to be particularly troublesome for certain NIPS platforms, and may be prone to causing erroneous suspicions of a deletion or a partial monosomy.
My point is, there is a good deal of data to suggest that NIPS is really excellent at screening for trisomy 21, and quite good at screening for trisomy 13 and trisomy 18 (although there is a whole separate discussion to be had on the positive predictive value of NIPS based on the incidence of the disorder it is intended to detect. See Katie Stoll’s previous post for an excellent discussion of that topic.) What I am concerned with is the practice of selectively reporting potential, but unvalidated abnormalities. As a genetic counselor, I much prefer that my patients have accurate information from a diagnostic test, rather than relying on a screening test, no matter how good of a screen it is. However, if I have a patient who is keen to avoid “invasive” testing, I would probably struggle with whether or not to tell her about an incidental finding of unclear validity, knowing that this information it may unduly influence her decision about whether or not to undergo diagnostic testing. Quite frankly, the onus should be on the laboratory NOT to disclose unvalidated results following an “official” interpretation of a result as non-reportable.
Fetal Aneuploidy or Maternal Malignancy?
Recently, the reporting of incidental, unvalidated results on NIPS received significant media attention when it came to light that NIPS results may serve as a marker for unrecognized maternal malignancies. BuzzFeed published an article detailing how an obstetrician who had ordered NIPS on one of his patients was shocked to learn that the laboratory suspected his patient may have cancer based on the results of her NIPS. While serendipitous discoveries such as this can lead to major improvements in diagnostic testing, it is not clear at which point this unvalidated information should be disclosed to the clinician. In this particular situation, it turned out well for the patient, who did, indeed, have cancer. However, if the information had been erroneous, and she did not have cancer, things could have taken a very different course. I think as clinicians, most of us recognize that incidental findings are a part of medicine. But this should be even more reason to determine how best to handle unanticipated results before they occur. Clear policies regarding what types of incidental findings can be/will be reported need to be put into place, and patients should be made aware of these possibilities as part of the informed consent process.
What Is the Solution?
I am all for making NIPS the best screening tool it can possibly be. I support prospective clinical trials to evaluate the tests’ performance and to determine what other abnormalities can reliably be detected. However, the reporting of analytically un-validated results, needs to stop. I understand the temptation to allow exciting preliminary data bleed over into clinical care, but when it comes down to it, we must commit ourselves to providing conscientious, responsible care. To me, this means rigorously validating and replicating results across heterogeneous populations before applying that information to the practice of medicine.