4 DISCUSSION
The prognostic nature of histological tumor necrosis in post-neoadjuvant chemotherapy surgical specimens in osteosarcoma has been emphasized by studies across different co-operative groups, where majority have dichotomized response based on an arbitrary 90% TN cut-off.3,6,8,13 Though good responders with ≥90%TN had better disease-free survival compared to poor responders, neither increasing the proportion of good responders nor intensifying treatment of poor responders has translated into better survival.3,10,14,15 This questions the predictive nature of this arbitrary cut-off for survival in management of osteosarcoma. Studies have also considered alternative cut-offs of 70%, 50% tumor necrosis as surrogate measures of outcome.11,12 The above studies question the value of this 90% arbitrary cut-off for TN on a MAP-based chemotherapy backbone. In this context, our study analyzed the prognostic significance of various already established TN cut-offs of 100% and 90% on a non-MAP based indigenous chemotherapy protocol. The relative distribution of the patients among the various cohorts with respect to above cut-offs of TN were similar to published data from western world, though the proportion of larger primary tumors (≥8cm) were higher in our study cohort (60% vs 45-49%), albeit difference in the definition of tumor size across the published studies (>150ml or >one-third of the involved bone or ≥8cm).8,16,11 In addition, data on 70% or alternative cut-offs which was shown to be prognostic in a very small cohort study is not available in these large cohorts.11
The prognostic significance for EFS persisted across all three cut-offs for whole and localized cohorts, with a gradual decline in EFS with decreasing necrosis. This is akin to some of the larger studies,2,3 though 90% cut-off failed to show prognostic value in some other studies.11,17 There was no differential impact on EFS of 100% and 90-99% TN in the above same studies.11,17 Moreover, all these MAP-chemotherapy based studies included patients above 15years as well (mostly patients <40years of age) and non-MAP St Jude study included only localized osteosarcoma.2 In our study, for the metastatic cohort though, 70% TN was prognostic for EFS with a HR of 3.41, and both 100% and 90% did not impact EFS. Overall Survival was affected only by 90% TN, in the whole and localized cohorts and none affected OS in metastatic cohort. This underlines the controversial role of tumor necrosis or histological response as a surrogate measure of outcome in the current defined status. Since our study found 70% TN to be prognostic for EFS across all groups, we decided to explore the optimum cut-off value of TN using ROC for these cohorts. A TN cut-off of 85.5% with an AUC of 69% and 70% respectively for whole and localized cohorts impacted EFS and in the metastatic cohort similar values were 84.5% TN at an AUC of 67%. This suggests the possibility of investigating better cut-offs of TN for predictive or prognostic value rather than using an arbitrary cut-off with no clinical meaning. This has better utility in the context of a non-MAP chemotherapy backbone to explore the survival incremental value with the addition of HD-MTX in an adjuvant setting for thus defined poor responders. Also, thus defined TN cut-off could help in triaging patients who are to be offered further treatment in the metastatic setting in resource constrained settings, where extremely poor prognosis is seen for poor responders.
The type of relapses seen in the cohort are similar to what is reported in western literature,18 mainly metastatic to lungs with no significant differences in the pattern of relapse based on tumor necrosis across the various subsets. This is in accord with what is already known in MAP-based chemotherapy studies.18 The study also looked at various clinical and laboratory parameters which predicted good and poor responders at various cut-offs of TN. Though a previous reported study has shown no predictive value for tumor necrosis amongst the various parameters,11 our study showed amputation across all cut-offs and male gender at a cut-off of 90% to predict poor necrosis, probably driven by bad biology.
This study has its own limitations in that it is a retrospective analysis with a relatively shorter follow-up and smaller sample size, especially in the setting of metastatic disease. Despite its shortcomings, this cohort represents a single center experience of a large group of a rare disease treated on a uniform non-MAP chemotherapy protocol. The availability of tumor necrosis data recorded as an absolute value permits for analysis at various cut-offs of TN compared to studies which has dichotomized the patients into good and poor responders based on a single cut-off of TN.
This analysis provides a basis to explore different cut-off levels for TN especially in LMICs with higher tumor burden and possible different biology for prognostic as well as predictive value in larger multicentric studies. Exploring alternative cut-offs with predictive or prognostic value would also help in tailored risk stratification of treatment approaches based on newer drugs or biologic agents in future studies.