Autism Research Community Enraged About Upcoming Release of Predictive Test
At best the test has been called “premature.” Others have suggested the test is “misleading,” built on “shaky data,” and will be inappropriately marketed to a desperately worried and motivated group of women. The test at the center of this firestorm is the MAR Autism test (Maternal Autoantibody-Related) developed by Pediatric Bioscience (PB; Sacramento, Calif.). […]
At best the test has been called “premature.” Others have suggested the test is “misleading,” built on “shaky data,” and will be inappropriately marketed to a desperately worried and motivated group of women. The test at the center of this firestorm is the MAR Autism test (Maternal Autoantibody-Related) developed by Pediatric Bioscience (PB; Sacramento, Calif.). The company, which declined a request to speak to DTET, claims MAR is “an informational test which determines if an individual has an increased chance of having a child with autism” and is intended for use as “a family planning tool” in women of child-bearing age who have already had a child with autism and child-bearing women over the age of 30 years, as well as in women of children 12 months to 24 months seeking a diagnosis for their child’s developmental delay. The test, the company says on its Web site, currently measures eight specific autoantibodies specific to fetal brain proteins that have been linked to the MAR form of autism. The test, which is expected to be commercially available in the second half of 2014, is based on work by a group of University of California, Davis researchers, including Judy Van de Water, Ph.D., who is chief scientific adviser to PB and licensed the patented antibody screening to the company. Their study, published in Translational Psychiatry this past July, says reactivity in a panel of antigen markers (lactate dehydrogenase A and B, cypin, stress-induced phosphoprotein 1, collapsin response mediator proteins 1 and 2, and Y-box-binding protein provides “99 percent specificity” of MAR autism risk. A slew of autism experts publicly dispute these findings, including in a news piece published in Science on Sept. 13. In addition to some skepticism about the suggested mechanism by which a direct antigen–antibody interaction inside the blood-brain barrier causes functional interference of the target proteins, researchers point to the study’s small sample size, lack of independent validation of the markers, and “weak” statistical analysis used to derive the study’s claims. The study concluded that nearly 23 percent of the 246 mothers of children with autism tested had one of the “specific combinations” of autoantibodies against the target proteins. Critics say that combining reactivity patterns, none of which were seen in more than 7 percent of mothers tested, exaggerates the test’s predictiveness. Steven Goodman, M.D., Ph.D., a clinical trial expert and Stanford University biostatistician, calculated for the Science article that the test has an actual positive predictive value (PPV) of 16.5 percent (meaning that only one out of six positive tests would be correct) and calls the company’s claim that a woman with a positive test has a 99 percent likelihood of having a child with autism “completely false.” “Seventy-seven percent of the time the test will miss a positive case and a positive test will be wrong 83 percent of the time—it is really bad both coming and going,” George Anderson, Ph.D., an autism researcher at Yale University tellsDTET. Anderson adds that there are other problematic autism tests emerging, including one that uses placental tissue, and despite the hope of a reliable test, the heterogeneous nature of autism spectrum disorder and its relatively low prevalence makes test development challenging. While it is tempting to be able to identify a specific cause or risk factor for autism (currently only 10 percent to 20 percent of cases can be tied to a wide variety of genetic mutations), Tom Wassink, M.D., from University of Iowa, tells DTET that at this point there are no markers—proteomic, genetic, or imaging—that are capable of diagnosing or predicting autism. “Folks affected by autism are desperate for an early identifier because early treatment makes a difference, but we are just not there yet,” Wassink acknowledges. “I don’t see the ability to predict autism with a biological test on the horizon, not in the next five years.” Takeaway: Despite hope that there will be a quantifiable means aside from a provider’s clinical acumen to detect autism early, there is currently insufficient evidence that genetic, biological, or imaging markers are able to identify those at risk for autism.