To critically assess the clinical utility of striational antibodies (StrAbs) within paraneoplastic and myasthenia gravis (MG) serologic evaluations.
All Mayo Clinic patients tested for StrAbs from January 1, 2012, to December 31, 2018, utilizing Mayo’s Unified Data Platform (UDP) were reviewed for neurologic diagnosis and cancer.
A total of 38,502 unique paraneoplastic evaluations and 1,899 patients with MG were tested. In paraneoplastic evaluations, the StrAbs positivity rate was higher in cancer vs without cancer (5% [321/6,775] vs 4% [1,154/31,727]; p < 0.0001; odds ratio [OR] 1.35; confidence interval [CI] 1.19–1.53), but receiver operating characteristic (ROC) analysis indicated no diagnostic accuracy in cancer (area under the ROC curve [AUC] 0.505). No neurologic phenotype was significantly associated with StrAbs in the paraneoplastic group. Positivity was more common in all MG cancers compared to paraneoplastic cancers (p < 0.0001). In MG evaluations, the StrAbs positivity rate was higher in those with cancer vs without (46% [217/474] vs 26% [372/1,425]; p < 0.0001; OR 2.39, CI 1.9–2.96), with ROC analysis indicating poor diagnostic accuracy for thymic cancer (AUC 0.634, recommended cutoff = 1:60, sensitivity = 56%, specificity = 71%), with worse accuracy for extrathymic cancers (AUC 0.543). In paraneoplastic or MG evaluations, the value of antibody positivity did not improve cancer predictions. Paraneoplastic evaluated patients with positive StrAbs were more likely to obtain CT (p = 0.0001), with cancer found in 3% (12/468).
Despite a statistically significant association with cancer, an expansive review of performance in clinical service demonstrates that StrAbs are neither specific nor sensitive in predicting malignancy or neurologic phenotypes. CT imaging is overutilized with positive StrAbs results. Removal of StrAbs from paraneoplastic or MG evaluations will improve the diagnostic characteristics of the current MG test.
Classification of Evidence
This study provides Class II evidence that the presence of StrAbs does not accurately identify patients with malignancy or neurologic phenotypes.