CAN SERT SCORE PREDICT HISTOLOGICAL RECURRENCE IN PIECEMEAL ENDOSCOPIC MUCOSAL RESECTION? – A COMPARATIVE STUDY WITH SMSA SCORE

Aims:

Piecemeal endoscopic mucosal resection (pEMR) allows the resection of larger non-invasive colorectal lesions. Prediction of histological recurrence (HR), as determined in the first surveillance colonoscopy, is still a limitation. The recently described SERT score (Sydney EMR-recurrence tool) can predict endoscopic recurrence (ER), however it was not evaluated as a predictor of HR. The present study aimed to validate the SERT score as a predictor of ER and HR as to compare it with the SMSA score (size, morphology, site, access) already validated as a predictor.

Methods:

The SERT and SMSA scores were calculated for all lesions submitted to pEMR between 2012 – 2018. In the first surveillance colonoscopy, performed at 3 – 6 months, ER and HR were evaluated. In the absence of ER, biopsy of the scar was performed in most cases. In patients with ER, removal of the residual lesion was attempted.

Results:

188 pEMR were considered. 61.7% were males and the mean age was 66.1 ± 9.8. Most lesions were located at the right colon (n = 47, 59.0%). Mean size was 31.0 ± 13.8 mm. ER was suspected in 27.1% (n = 51), and in 72.5% of these cases (n = 37) HR was confirmed.

There was a moderate positive correlation between the SERT and SMSA scores (p < 0.001;r = 0.57). There was a significant association between SMSA score and ER (p < 0.001) as well as HR (p < 0.001). SERT score was also significantly associated with ER (p = 0.005) and HR (p = 0.015). The SERT score showed a discriminative power to ER and HR, not significantly different from SMSA score (p = 0.5).

Conclusions:

The SERT score correlates with the SMSA score and both can be used to predict RE and HR in lesions removed by pEMR. The SERT score showed a similar discriminative power for endoscopic and histologic recurrence. The SERT score is less complex and allows predicting not only endoscopic recurrence but histologic recurrence as well.