Validation of a novel risk score to predict early and late recurrence in solitary fibrous tumour

BACKGROUND

Current risk models in solitary fibrous tumour (SFT) were developed using cohorts with short follow-up and cannot reliably identify low-risk patients. We recently developed a novel risk model (G-score) to account for both early and late recurrences. Here, we aimed to validate the G-score in a large international cohort with long-term follow-up.

METHODS

Data were collected from nine sarcoma referral centres worldwide. Recurrence-free interval (RFi) was the primary endpoint.

RESULTS

The cohort comprised 318 patients with localised extrameningeal SFTs. Disease recurrence occurred in 96 patients (33%). The estimated 5-year RFi rate was 72%, and the 10-year RFi rate was 52%. G-score precisely predicted recurrence risk with estimated 10-year RFi rate of 84% in low risk, 54% in intermediate risk and 36% in high risk (p < 0.001; C-index 0.691). The mDemicco (p < 0.001; C-index 0.749) and SalasOS (p < 0.001; C-index 0.674) models also predicted RFi but identified low-risk patients less accurate with 10-year RFi rates of 72% and 70%, respectively.

CONCLUSIONS

G-score is a highly significant predictor of early and late recurrence in SFT and is superior to other models to predict patients at low risk of relapse. A less intensive follow-up schedule could be considered for patients at low recurrence risk according to G-score.

A novel risk score to predict early and late recurrence in solitary fibrous tumour

AIMS

Solitary fibrous tumours (SFTs) are rare mesenchymal neoplasms with recurrence rates of 10-30%. Current risk stratification systems for extrameningeal SFTs are based on cohorts with limited follow-up and are not suitable for prediction of late recurrences. In this study we aimed to develop a prognostic model accounting for both early and late recurrences using a relatively large patient cohort with long-term follow-up.

METHODS AND RESULTS

Clinicopathological factors were analysed in a cohort of 100 extrameningeal, STAT6-positive SFTs. Median follow-up for overall survival (OS) and recurrence-free interval (RFi) were 121 and 84 months, respectively. Disease relapse occurred in 31% of patients and median time to recurrence was 63 months. In univariate analysis mitotic count, necrosis, male gender and presence of severe atypia and pleomorphism were associated with inferior RFi. Mitotic count, necrosis and male gender were independent predictors of recurrence in multivariate analysis. Previously published risk models were also statistically associated with RFi in our cohort, but failed to reliably identify low-risk patients due to poor prediction of late recurrences. A novel risk score based on mitotic count, necrosis and gender was able to stratify patients into low-, intermediate- and high-risk groups for both early and late recurrences.

CONCLUSIONS

In this cohort of patients with extrameningeal SFT and long-term follow-up mitotic count, necrosis and gender were independent prognostic markers of recurrence. We propose a novel risk score based on these factors and accounting for late recurrences, which should be validated in external cohorts with sufficient follow-up time.

Clinical and molecular implications of NAB2-STAT6 fusion variants in solitary fibrous tumour

Solitary fibrous tumour (SFT) is a mesenchymal neoplasm characterised by pathognomonic NAB2-STAT6 gene fusions. The clinical implications and prognostic value of different fusion variants has not been clarified. In the current study, we explore the clinicopathological, prognostic and molecular differences between tumours with different fusions. Thirty-nine patients with localised, extrameningeal SFT were included, of whom 20 developed distant recurrence and 19 were without recurrence after long term follow-up. Capture-based RNA sequencing identified 12 breakpoint variants, which were categorised into two groups based on the STAT6 domain composition in the predicted chimeric proteins. Twenty-one of 34 (62%) sequenced tumours had fusions with most of the STAT6 domains intact and were classified as STAT6-Full. Thirteen tumours (38%) contained only the transactivation domain of STAT6 and were classified as STAT6-TAD. Tumours with STAT6-TAD fusions had a higher mitotic count (p=0.016) and were associated with inferior recurrence-free interval (p=0.004) and overall survival (p=0.012). Estimated 10-year recurrence-free survival was 25% for patients with STAT6-TAD tumours compared to 78% for the STAT6-Full group. Distinct transcriptional signatures between the fusion groups were identified, including higher expression of FGF2 in the STAT6-TAD group and IGF2, EGR2, PDGFRB, STAT6 and several extracellular matrix genes in STAT6-Full tumours. In summary, we demonstrate that NAB2-STAT6 fusion variants are associated with distinct clinicopathological and molecular characteristics and have prognostic significance in extrameningeal SFT.