International Cancer Genome Consortium Highlights Inter-Lab Variations in Whole Genome Sequencing

Even as whole genome sequencing (WGS) is replacing whole exome sequencing in cancer research and is being adopted in clinical practice, scientists have been aware of variations in the quality of WGS. New research shows just how wide the inter-laboratory gap is for calling somatic mutations.

A team of International Cancer Genome Consortium (ICGC) researchers led by the Spanish Centro Nacional de Analisis Genómico (CNAG) and the German Cancer Research Center found significant issues both in sequencing and evaluating WGS results (Nat Commun 2015; doi:10.1038/ncomms10001).

Using chronic lymphocytic leukemia (CLL) and medulloblastoma tumor samples, the investigators found that the participating institutions identified only 40% of more than 1,000 confirmed somatic single-base mutations. Moreover, participants identified just one of 337 single somatic insertion/deletion mutations (0.3%).

The findings underscore the need to improve cancer WGS, according to senior author, Ivo Gut, director of CNAG. “The findings of our study have far-reaching implications for cancer genome analysis,” he explained. “We have found many inconsistencies in both the sequencing of cancer genomes and the data analysis at different sites. We are making our findings available to the scientific and diagnostic community so that they can improve their systems and generate more standardized and consistent results.”

In contrast to other studies that have addressed WGS quality issues using simulated tumor genomes, this investigation involved real tumor-normal pairs.

The project team provided participating laboratories with a set of unaligned WGS reads of CLL of average quality at 40X coverage. They also provided matched samples of medulloblastoma and its corresponding normal genome (sequenced at 40X and 30X coverage, respectively). ICGC labs then returned their somatic mutation calls.

The researchers went on to merge the sequencing data from each ICGC lab to yield a benchmarked deep sequence up to 300X coverage on the tumor-normal pairs and have made this curated set of verified somatic mutations available for download.

Based upon their findings, the investigators developed an eight-point checklist for WGS cancer studies, which calls for PCR-free library preparation protocols to ensure evenness of coverage, and a sequencing depth close to 100X coverage for both tumor and normal specimens.

No Advantage to One-Step Gestational Diabetes Screening Method

Changing to a one-step gestational diabetes screening algorithm from a two-step process increased the percentage of women identified with gestational diabetes but had no significant effect on the rate of large for gestational age (LGA) babies (Obstet Gynecol 2016;0:1–8). The findings do not support transitioning to the one-step approach, according to the authors.

Throughout much of the United States and as recommended by the American Congress of Obstetricians and Gynecologists, the standard gestational diabetes screening practice involves two steps, an initial 50-g glucose screen followed by a 3-hour diagnostic oral glucose tolerance test using either Carpenter-Coustan criteria or National Diabetes Data Group criteria. However, in 2010 the International Association of the Diabetes and Pregnancy Study Groups (IADPSG) and the World Health Organization recommended first identifying undiagnosed pregestational diabetics at their initial prenatal visit using a hemoglobin A1c test, a random plasma glucose test, or a fasting plasma glucose test. Women not identified as having diabetes in this step would then, at between 24 and 28 weeks’ gestation, undergo a one-step 2-hour glucose tolerance test with a 75-g glucose load.

The study authors sought to compare outcomes associated with these two screening approaches after their hospital switched from the two-step Carpenter-Coustan method to the one-step IADPSG approach. Their retrospective before-and-after cohort study involved a total of 6,828 patients.

The primary outcome was LGA, defined as a birth weight greater than the 90th percentile for the baby’s gestational age and sex. Secondary outcomes included intensive care unit admission, preterm delivery, preeclampsia, and hyperbilirubinemia.

The authors found that 27% of patients were diagnosed with gestational diabetes under IADPSG criteria versus 17% using Carpenter-Coustan. However, there was no significant difference in rates of LGA between the two approaches (9% under IADPSG versus 10% under Carpenter-Coustan) nor in secondary outcomes. Of note, the investigators found a significantly higher primary cesarean delivery rate in the IADSPG group, even after women without gestational diabetes were included (20% versus 16%).

The latter finding suggests that the diagnosis of gestational diabetes by itself “may lower the threshold for cesarean delivery,” according to an accompanying editorial. The authors suggested that IADPSG cutoffs might be too low, resulting in “too many patients being treated as having gestational diabetes” and they called for further studies to evaluate different thresholds.

Thyroglobulin-Galectin-3 Binding Could Confound Testing Methods

Michigan Technological University researchers recently reported that thyroglobulin (Tg) and galectin-3 (Gal-3) form a complex through high affinity binding that could interfere with both Tg and Gal-3 assays and impact interpretation of these tests (Thyroid 2015;25:1–8).

The researchers used hemagglutination inhibition assays and isothermal titration calorimetry to demonstrate that Gal-3 binds with high affinity to Tg. They also found that Gal-3 has “enhanced affinity” for poly-LacN-acetyllactosamine (LacNAc) chains in comparison with LacNAc. Since “Tg produced by transformed thyroid cells contains poly-LacNAc units in their complex-type glycans” the investigators speculated that Gal-3 binds to these cells with “even higher affinity.”

Gal-3’s high affinity binding to Tg could produce an interference problem in testing similar to that produced when Tb autoantibodies (TgAb) bind to Tg. This binding also sequesters both Tg and Gal-3, decreasing amounts of both free Tg and free Gal-3.

While it takes about 24 hours to form Tg-TgAb complexes, the Tg-Gal-3 complex forms within minutes, according to the authors. They found that both free Tg and Gal-3 dissolve Tg-Gal-3 complexes; however, Gal-3 does so much less efficiently.

Heart Failure Treatment Rarely Monitored as Recommended

Patients taking mineralocorticoid receptor antagonists (MRAs) for heart failure rarely receive guideline-recommended early or extended lab test monitoring, new research indicates (JAMA 2015;314:1973–5).

MRAs like spironolactone and eplerenone are commonly prescribed for patients with heart failure, but American College of Cardiology/American Heart Association guidelines recommend close monitoring of patients’ renal function and electrolyte levels because MRAs carry a risk of hyperkalemia.

The researchers accessed Centers for Medicare and Medicaid Services data for more than 10,000 Medicare beneficiaries to assess how well patients’ serum creatinine and potassium levels were assessed both before and during MRA therapy. They considered appropriate MRA monitoring as creatinine or potassium testing 120 days before the patient started taking MRAs. They also looked for two or more measurements within 1 to 10 days of starting MRAs, and at least three measurements between days 11 through 90 after starting MRAs.

While 92% of patients had testing before starting on MRAs, just 13% had appropriate testing done within 10 days of initiating MRA therapy, and only 30% received appropriate extended monitoring.