Grade Missingness Does Not Appear to Affect Outcomes in Pediatric Surgical Oncology Patients

Background: The impact of grade on prognosis is highly variable and is inconsistently included in the pediatric surgical oncology literature. Contrastingly, in the adult surgical oncology literature, grade has been noted to be missing more frequently and oncologic outcomes are worse in under-resourced communities. Therefore, the aim of this project is determine the association – if any – between missingness of histologic grade and demographic, clinical, or hospital factors for pediatric surgical oncology patients and to examine the association between grade missingness and outcomes after reliability-adjusting hospitals for rate of grade missingness.

Methods: The National Cancer Database (NCDB) was queried for all patients <21 years with the most common single organ solid tumor malignancies: Wilms tumor (n=2,680) and hepatoblastoma (n=475). Missingness of histologic grade was recorded. Multivariable logistic regression was used to determine the association between demographic and clinical factors and missingness of histologic grade. After reliability adjustment, hospitals were ranked in terms of rate of missingness for each tumor type, and patient factors, treatment, and unadjusted survival were compared for patients across hospital quintiles.

Results: Only 38/475 patients with hepatoblastoma and 199/2,680 patients with Wilms had grade data present. For both tumor types, missingness of histologic grade was not associated with age category, race, insurance status, or income quartile. Similarly, after reliability-adjustment of hospitals by rate of missingness, there was no difference in the distributions of patients’ demographic and clinical factors in hospitals in the highest and lowest quintiles of missingness. There were no differences in tumor size, metastases at diagnosis, or oncologic therapy, with all patients undergoing surgical resection (Tables 1 and 2). Finally, the median survival [95% confidence interval] was not significantly different between patients with grade present or absent (66 months [30-87] vs 62 months [55-65], p=0.742).

Conclusion: Large national databases have become critical for studying rare tumors in pediatric patients. In contrast to studies of databases in adult oncology patients, our findings suggest that grade missingness in pediatric surgical patients is likely at random and not associated with demographic or clinical factors on presentation. Furthermore, grade missingness is not associated with treatment paradigms or survival outcomes at the hospital level. Given the sparseness of data even within national databases in pediatric solid tumors, this data provides a foundation for including patients with incomplete histology data in surgical studies.