A new study has found that oncology centers frequently make discordant diagnoses for myelodysplastic syndromes (MDS). The findings have implications not only for patients but also for the accuracy of study-related and national registries, according to Edward Gorak, DO, MBA, MS, a hematologist/oncologist at Baptist MD Anderson Cancer Center in Jacksonville, Florida, and colleagues. The research, which was published in Blood Advances, suggests that many patients may be receiving inappropriate treatment.
“This is kind of a big deal… and was both surprising and not surprising,” said Mikkael A. Sekeres, MD, MS, chief of hematology at the University of Miami Sylvester Cancer Center in Florida and senior author of the paper.
The study included 918 patients who enrolled in the ongoing National Heart, Lung, and Blood Institute National MDS Natural History Study, a prospective observational cohort study of participants with suspected MDS or MDS/myeloproliferative neoplasms (MPN). Each patient underwent a bone marrow biopsy and obtained a diagnosis from a local pathologist at one of more than 140 sites across the U.S. The bone marrow sample was then sent to two central experts, and if those experts disagreed on the diagnosis, it was sent to a third expert.
Approximately one-third of cases were reclassified (given a different diagnosis) following central review: 29% MDS (n=266), 5% MDS/MPN overlap (n=45), 5% idiopathic cytopenia of undetermined significance (ICUS; n=49), 2% acute myeloid leukemia with less than 30% blasts (n=15), and 59% other (n=543). “We expected some discrepancies, but we didn’t expect it to be that much,” Dr. Sekeres said.
The authors noted in their discussion that the study was not designed to determine any one site’s diagnostic accuracy, and because it was a U.S. study, the findings may not be generalizable internationally.
After they identified the high rate of discordance, the researchers sought to determine the source of the discrepancy. They found that a main driver was that research coordinators at each of the study sites were frequently entering the wrong diagnoses into the registry. As is typical for such registries, Dr. Sekeres explained, staff enter data manually and choose from a drop-down menu of diagnoses. In the current study, site miscoding errors accounted for 53% of the local misdiagnoses, which left a true misdiagnosis rate of 15% overall and 21% for MDS. The researchers then examined data from the patients who received a true misdiagnosis and asked how many of them received inappropriate therapy. They found that 7% of patients who were misdiagnosed received the wrong therapy.
“This is a problem,” Dr. Sekeres said. “Pathologists get it wrong, and if the pathologists I work with are no good, then I’m no good. Fortunately, the pathologists I do work with are very good!”
He suggested that patients, especially those diagnosed with a rare hematologic malignancy, obtain a second opinion. He explained that this may be particularly important for patients seeking a diagnosis at a general hospital because pathologists at general hospitals may not have a great deal of experience with diagnosing MDS or leukemia, especially given that the classification system for MDS and leukemias has become more complicated and that successful therapies have become more targeted to specific diagnoses.
“What about all those patients where it is just a keystroke error?” Dr. Sekeres added. As the lead investigator in the MDS Natural History Study, he is concerned about the implications for not just his study but also other national cancer registries or clinical trials that often have diagnoses entered by hand. Such keystroke errors could theoretically be undermining all national cancer registries.
Dr. Sekeres explained that, in the case of clinical trials, coordinators at local centers do submit data to the U.S. Food and Drug Administration (FDA) for review. “The FDA does audit clinical trials,” Dr. Sekeres said, “but it does not audit every patient enrolled in a clinical trial.” He thus proposed that data be entered into national registries by people who are specialists in the disease for which they are entering data. Those data should then be checked for accuracy before they are submitted to national registries.
Any conflicts of interest declared by the authors can be found in the original article.
Gorak E, Otterstatter M, Al Baghdadi T, et al. Discordant pathological diagnoses of myelodysplastic neoplasms and their implications for registries and therapies [published online ahead of print, 2023 Aug 8]. Blood Adv. doi: 10.1182/bloodadvances.2023010061.