The Daily Sign-Out

FDA draft guidelines on IVD "companion diagnostics"

Digital Pathology Insights blog posted a nice succinct comment on the FDA draft guidance on companion diagnostics that are of acute interest for anyone interested in digital pathology.

The issue is the development of drugs (i.e., so-called "targeted" drug therapies, such as, trastuzumab) that depend on the result of a predictive test (HER2 protein overexpression by IHC, amplification by FISH) to meet the claims of the drug's effectiveness as approved by the FDA.  As the FDA document states in the introduction,

When an appropriate scientific rationale supports such an approach, FDA encourages the development of therapeutic products that depend on the use of approved or cleared IVD companion diagnostic devices — several such IVD companion diagnostic devices for use with corresponding therapeutic products have already been approved or cleared.

As pointed out in the blog post, the implication is that the drug will not get approval unless the companion diagnostic test or methodology gets approval (and vice versa?).

You may not want to read the FDA document before operating heavy machinery or driving.  

Incidentalomas revisited

Dr. Bruce Friedman posted a thoughtful blog on Lab Soft News recently concerning "incidentalomas" ("How to Avoid the Risks of a CT Incidentaloma").  This post happened to coincide with a new working committee at my hospital commissioned with developing a virtual clinic for following patients who have incidental lung nodules discovered.

Perhaps unlike other sites, there are widely disseminated radiographic recommendations for following lung nodules (Fleishner criteria).  Of interest, a recent article found that despite widespread awareness of these criteria, actual incorporation of these criteria into practice has been found wanting--only around 50% of radiology practices actually used these recommendations in reports.  One issue that we will be challenged with is how to track these patients across multiple different providers for appropriate follow-up.  One model we are considering is already in place: using a breast "navigator" to follow-up patients with abnormal mammograms.  This issue arose recently because of a series of patients presented at our Tumor Board who had incidental lung nodules identified but were not followed appropriately and then were re-discovered with inoperable advanced stage disease months later.

This idea also came up just today in a conversation I was having with a urologist colleague regarding renal incidentalomas treated with cryoablation.  I recently diagnosed an incidentally discovered 2.5-cm mass in an 80-year-old woman on CT-guided needle biopsy as an muscle-predominant angiomyolipoma (mini-case report on its way!).  For better or worse, the tumor was ablated probable before it was even grossed!  But, to return to Dr. Friedman's post, what if the patient was simply observed with serial imaging?  The urologist was unaware of any recommendations similar to the Fleishner criteria for lung nodules.  If you, dear reader, know of any please post them in a comment.

Finally, today's email brought a "Published Ahead of Print" article from Journal of Thoracic Oncology on-- mediastinal incidentalomas.  This is a prospective study of 88 patients in which mediastinal lymphadenopathy was found as an incidental finding on CT scans for indications other than staging.  The title is a bit deceptive but the scenario is common.  The finding of incidental mediastinal lymph nodes on CT were characterized by multiple Naruke stations involved, relative small sizes of nodes (median ranging from 6 to 14 mm for all nodes), and coexistence with hilar adenopathy.  In this study, sarcoidosis was diagnosed by EBUS-FNA in 22% of patients and 2 patients developed lung cancer 2 years after initial finding.

While I concur with the basic premise of the post (i.e. that one shouldn't jump into a full-blown diagnostic work-up of an incidental finding), I think we should also insist on developing evidence-based criteria for managing these types of findings such as, Fleishner criteria for lung nodules.

CDC: Discordant results from "reverse sequence" syphilis screening

Currently, the CDC recommends screening with a nontreponemal test, such as the RPR, to identify possible patients with untreated syphilis, followed by confirmation with one of several treponemal tests.  However, after the introduction of automated treponemal enzyme/chemiluminescence assays (EIA/CIA), some labs (like mine) adopted a "reverse sequence" screening strategy of a screening treponemal EIA/CIA, with positive sera followed with a nontreponemal test.  The CDC reported last week in MMWR the results of a study of 140,176 sera screened with treponemal EIA/CIA tests from five laboratories, including three sites serving patient populations with a low disease prevalence and two sites serving patient populations with a high disease prevalence, that have used "reverse sequence screening" for syphilis from 2006-2010.

among sera reactive on initial screening with a treponemal EIA/CIA, 56.7% had a nonreactive RPR test. Among these discordant sera, 31.6% also were nonreactive by treponemal testing using Treponema pallidum particle agglutination (TP-PA) or fluorescent treponemal antibody absorbed (FTA-ABS) tests. Among discordant sera, the rate of nonreactive confirmatory treponemal tests was 2.9 times higher in a population with low prevalence of syphilis,(my emphasis) suggesting that the low-prevalence population had a higher percentage of false-positive test results.

This certainly reflects my experience as well.  It is awkward to explain both the rationale and the discordant results to clinicians, most of whom must take time from their schedule to call one of the pathologists to explain the results.  Moreover, most of these tests are ordered on healthy women as part of a pregnancy screening panel--exactly the low prevalence population identified above by the CDC as having a higher incidence of false-positive results.  Although we explain that the discordant results are probably falsely positive, but we also add that the results also could be either from a previous syphilis infection in which the nontreponemal antibodies are no longer detected or from early primary syphilis before nontreponemal antibodies have developed--not so reassuring to the clinician.  They invariably ask, "Now what?"  Basically, we dump it back on them.  Discrepant results cause undue anxiety for physicians and patients and creates nagging doubts about other lab results, the lab, the staff.  On the other hand, the lab staff likes the automation of the EIA test because the RPR is a time-intensive, labor-intensive manual test.

The CDC is still recommending screening with a nontreponemal test (RPR or VDRL), followed by a treponemal test, such as TP-PA or FTA-ABS, for reactive tests.  The article quite rightly reminds us of the importance of obtaining a thorough history and physical examination with regard to identifying high risk sexual behaviors, prior treatment for syphilis, and evidence of primary infection.  However, there are also recommendations and an algorithm for those labs using reverse sequence screening.  In brief, discordant results should be followed by TP-PA (preferably over FTA-ABS); positive TP-PA should be regarded as diagnostic for past or present syphilis, while negative TP-PA should be regarded as "unlikely."

This is essential reading if your lab performs syphilis testing or you occasionally are called to explain testing results.  The editorial note also nicely outlines future directions and research questions regarding syphilis testing.

Analyzing Malpractice Claims to Reduce Errors--a pathologist's perspective

The Informed Patient column by Laura Landro from yesterday's Wall Street Journal (September 27, 2010) has an interesting report entitled, "What the Doctor Missed" which discusses how analysis of malpractice claims is informing efforts to improve medical practice.

The issue of diagnostic errors due to missed or delayed diagnoses should be of particular interest to pathologists.  The article doesn't explicitly state this but poor communication (patient-to-physician, physician-to-patient, PCP-to-physician/specialist, etc.) seems to me a major factor in lawsuits and a major contributing factor in diagnostic errors.  How can we more effectively communicate our findings to non-pathologists?  I especially note reference the article makes to breast cancer.  We routinely use histological diagnostic terms--that can have crucial implications for patient care with respect to appropriate follow-up and risk assessment--which can bewilder even experienced surgical oncologists and radiologists.  An example from breast cancer would be communicating the spectrum of "columnar cell lesions."  If we identify this in a core biopsy for a mammographically-detected abnormality, do we routinely communicate any assessment of risk for concomitant or future cancer?  Is that the responsibility of the pathologist?  In a broader view, what is our responsibility in ensuring that physicians "receive" actionable lab results?  Can we design systems that can prevent errors "of omission" that result from missing critical lab information within a critical period?  A blogpost isn't sufficient space to devote to an in-depth discussion of these questions but the problem is more complicated than just mere documentation.

This is just some appetizers for consideration.  I wonder if one of the subjects we are missing in medical school might be communication (i.e., rhetoric and semiotics)--not just making PowerPoint presentations but I'm talking about how to talk to people (all different kinds of people), how to assess what they've heard or understood, and (perhaps most critically) how to listen.

European Congress of Pathology annual meeting

Clipped from The Dark Daily report:

"Next to speak was Bent Ejlertsen, M.D., Ph.D., of the Department of Oncology, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark. The title of his presentation was “Prediction of Responsiveness to Adjuvant Anthracyclines in High-Risk Breast Cancer Patients.”

He reported on the extensive studies about the TOP2A marker that he and his colleagues have conducted. Their goal was to learn how to use this biomarker for both prognostic and predictive evaluations to identify which breast cancer patients would benefit from therapy with adjuvant anthracyclines. The TOP2A marker is scored in a similar fashion as the HER2 marker. Ejlertsen stressed that improved patient safety would result as the pathology profession achieved more standardized processing and diagnosis when working with the TOP2A marker.

Third speaker at this special seminar on histopathology workflow and patient safety was Professor Giuseppe Viale, M.D., of the University of Milan School of Medicine and European Institute of Oncology in Milan, Italy. Viale emphasized the importance of achieving improved standardization for estrogen receptor (ER) and progesterone receptor (PgR) assays in breast cancer. Of particular note, Viale is a member of the international committee working to develop recommendations for standardizing the preparation and evaluation of tests utilizing ER and PgR immunohistochemical probes.

He opened his presentation by describing the importance of achieving improved standardization in ER and PgR testing. Viale stated that assessment of hormone receptor status is “the first and most important clue” in tailoring treatment of a breast cancer patient. Additionally, he noted that the importance of avoiding false positive results for this testing, because Tamoxifen is detrimental for patients with nonendocrine-responsive tumors."

My comments:

The topoIIA story is still not settled.  Most studies have not demonstrated the utility of this marker as a predictive or prognostic marker.

ER testing recommendations/standards are supposedly forthcoming from ASCO/CAP.  Participation in CAP Proficiency Testing or some other external validation program is a minimum.  As I've stated previously, attention to preanalytical variables that affect IHC staining as well as testing protocols, especially with respect to antigen retrieval, are crucial and yet, curiously, seem to be not taken seriously.  Choosing a control that has been externally validated to show negative, weak, and positive staining increased my confidence in our process and really helped "dial-in" our results with CAP-PT surveys.