乌鸦传媒

BALTIMORE -- The number of publications on sarcoma research that used registry databases has increased substantially over the last few years, but that doesn't necessarily mean that all of these studies have the same level of scientific rigor.

While the number of annual publications on musculoskeletal sarcoma that used data from the Surveillance, Epidemiology, and End Results (SEER) database and the National Cancer Database (NCDB) was stable from 2011 to 2017, it has since tripled in the last 3 years, reported Joshua M. Lawrenz, MD, of Vanderbilt University Medical Center in Nashville.

However, an evaluation of these publications suggested that "caution should be used when assessing and interpreting certain sarcoma data points derived from large national databases," Lawrenz said during a presentation at the annual Musculoskeletal Tumor Society meeting.

Of the 177 papers on sarcoma published over the last 3 years that used the two databases, 80% used SEER. These publications came from 14 different countries, with most from the U.S., followed by China, and from 119 total institutions, 53 of which had more than one publication and four with more than 10 publications.

Most papers were published in Cancer, Clinical Orthopaedics and Related Research, and the Journal of Surgical Oncology.

The most common type of paper was an overview paper, or the summary of a condition. Other topics included treatment, demographics, tumor characteristics, and surveillance. Treatment papers focused on adjuvant radiation, surgical resection type, and survival, while the most common type of tumor-focused papers involved risk factors for metastases, significance of lymph node involvement, and assessment of staging classifications.

A major issue with many of the studies that used these databases was that they often generated disparate outcomes, said Lawrenz. "We identified about a dozen topics in which different papers using the exact same database, during the same time frame, came to a different result."

For example, regarding the use of radiation in localized Ewing sarcoma and survival, Lawrenz noted that three papers -- all using SEER and published from 2016 to 2019 -- came up with different results, with one showing noninferior results with radiation compared with surgery, one showing reduced overall survival with radiation compared with surgery, and one showing no difference in survival between the two treatments.

In another case, two studies looking at retroperitoneal soft tissue sarcomas produced disparate overall survival rates of 59% and 70%.

Furthermore, Lawrenz pointed out that the abstraction error in these publications increased as oncologic details became "more nuanced." The researchers performed a coding concordance audit between their institution's administrative database and a prospective physician-kept database, which showed a high mismatch rate when looking at "nuanced factors," such as margin status (38%) and clinical stage (77%).

"Prior, larger, and similar coding audits in sarcoma have been performed and show up to a 40% discordance of administrative or billing data with medical records," Lawrenz said. "One hypothesis for this is that sarcoma diagnoses have a lot of increased heterogeneity, and this lack of organ specificity -- for instance, a breast angiosarcoma being incorrectly coded as a breast cancer -- can lead and contribute to why coding in sarcoma can be quite challenging."

"I think we would all agree that NCDB and SEER are well established, reputable, undergo stringent quality reviews, have the advantage of capturing incidence and population data, and can achieve statistical power when studying a rare disease like sarcoma," Lawrenz noted. However, he added that caution should be used when evaluating data points such as margins, clinical stage, or surgical intention.

While the launch of the American Academy of Orthopaedic Surgeons/Musculoskeletal Tumor Society musculoskeletal tumor registry provides an opportunity to improve in these areas, Lawrenz suggested that "we have to acknowledge we remain vulnerable to the same shortcomings in data collection unless a proactive effort is taken by clinicians to minimize them."

During a discussion after Lawrenz's presentation, Michelle Ghert, MD, of McMaster University in Hamilton, Ontario, suggested that one of the reasons for the recent increase in sarcoma publications is the proliferation of "paper mills" -- paid services that provide made-to-order scientific papers.

"This is something we should all be aware of," she said. "There are companies in certain countries that produce papers that are then sold to authors for publication. It's polluting the literature. We as a community need to recognize that this does happen."

Comment