Can histologic features predict neoadjuvant therapy response in rectal adenocarcinoma?

Tumour budding as a prognostic biomarker in biopsies and resections of neoadjuvant-treated rectal adenocarcinoma

BACKGROUND

Tumour budding (TB) is a marker of tumour aggressiveness which, when measured in rectal cancer resection specimens, predicts worse outcomes and response to neoadjuvant therapy. We investigated the utility of TB assessment in the setting of neoadjuvant treatment.

METHODS AND RESULTS

A single-centre, retrospective cohort study was conducted. TB was assessed using the hot-spot International Tumour Budding Consortium (ITBCC) method and classified by the revised ITBCC criteria. Haematoxylin and eosin (H&E) and AE1/AE3 cytokeratin (CK) stains for ITB (intratumoural budding) in biopsies with PTB (peritumoural budding) and ITB (intratumoural budding) in resection specimens were compared. Logistic regression assessed budding as predictors of lymph node metastasis (LNM). Cox regression and Kaplan-Meier analyses investigated their utility as a predictor of disease-free (DFS) and overall (OS) survival. A total of 146 patients were included; 91 were male (62.3%). Thirty-seven cases (25.3%) had ITB on H&E and 79 (54.1%) had ITB on CK assessment of biopsy tissue. In univariable analysis, H&E ITB [odds (OR) = 2.709, 95% confidence interval (CI) = 1.261-5.822, P = 0.011] and CK ITB (OR = 2.165, 95% CI = 1.076-4.357, P = 0.030) predicted LNM. Biopsy-assessed H&E ITB (OR = 2.749, 95% CI = 1.258-6.528, P = 0.022) was an independent predictor of LNM. In Kaplan-Meier analysis, ITB identified on biopsy was associated with worse OS (H&E, P = 0.003, CK: P = 0.009) and DFS (H&E, P = 0.012; CK, P = 0.045). In resection specimens, CK PTB was associated with worse OS (P = 0.047), and both CK PTB and ITB with worse DFS (PTB, P = 0.014; ITB: P = 0.019). In multivariable analysis H&E ITB predicted OS (HR = 2.930, 95% CI = 1.261-6.809) and DFS (HR = 2.072, 95% CI = 1.031-4.164). CK PTB grading on resection also independently predicted OS (HR = 3.417, 95% CI = 1.45-8.053, P = 0.005).

CONCLUSION

Assessment of TB using H&E and CK may be feasible in rectal cancer biopsy and post-neoadjuvant therapy-treated resection specimens and is associated with LNM and worse survival outcomes. Future management strategies for rectal cancer might be tailored to incorporate these findings.

Deep learning of endoscopic features for the assessment of neoadjuvant therapy response in locally advanced rectal cancer

BACKGROUND

For locally advanced rectal cancer (LARC), accurate response evaluation is necessary to select complete responders after neoadjuvant therapy (NAT) for a watch-and-wait (W&W) strategy. Algorithms based on deep learning have shown great value in medical image analyses. Here we used deep learning algorithms of endoscopic images for the assessment of NAT response in LARC.

METHOD

214 LARC patients were retrospectively included in the study. After NAT, these patients underwent total mesorectal excision (TME) surgery. Among them, 51 (23.8%) of the patients achieved a pathological complete response (pCR). 160 patients from Shanghai Changzheng Hospital were regarded as primary dataset, and the other 54 patients from Zhejiang Cancer Hospital were regarded as validation dataset. ResNet-18 and DenseNet-121 were applied to train the models based on endoscopic images after NAT. Deep learning models were valid in the validation dataset and compared to manual method.

RESULTS

The performances were comparable in AUC between deep learning models and manual method. For mean metrics, sensitivity (0.750 vs. 0.417) and AUC (0.716 vs. 0.601) in ResNet-18 deep learning model were higher than those in the manual method. The deep learning models were able to identify the endoscopic features associated with NAT response by the heatmaps. A diagnostic flow diagram which integrated the deep learning model to assist the clinicians in making decisions for W&W strategy was constructed.

CONCLUSIONS

We created deep learning models using endoscopic features for assessment of NAT in LARC. The deep learning models achieved modest accuracies and performed comparably to manual method.

Machine Learning of Histomorphological Features Predict Response to Neoadjuvant Therapy in Locally Advanced Rectal Cancer

AIM

We hypothesize that machine learning of histomorphological features can predict response to neoadjuvant therapy (NAT) in locally advanced rectal cancer (LARC).

METHOD

This retrospective study included 146 LARC patients who received NAT followed by surgery. The pathologists scanned the H&E slides of pretreatment tumor biopsy into whole slide images (WSIs). We randomly split patients into the primary and validation sets with a ratio of 80%:20%. We cut the WSIs into smaller parts (sample amount: 200-500) and used a convolutional neural network (CNN) to process these blocks directly. Then, a graph neural network (GNN) was applied to train the model in the primary set. The independent validation set was used to assess the performance of the model.

RESULT

Our model could provide indicative information to identify the patients who were most likely to benefit from NAT. When the sample amount reached 500, the tile-level classifier for distinguishing poor response from good response produced an AUC of 0.779 in the primary set and 0.733 in the validation set.

CONCLUSION

In this pilot study, we propose a novel predictive model of therapeutic response to NAT in LARC using a routine diagnostic tool employed in daily practice.

A novel prediction model for pathological complete response based on clinical and blood parameters in locally advanced rectal cancer

Background

The aim of this study was to investigate whether clinical and blood parameters can be used for predicting pathological complete response (pCR) to neoadjuvant chemoradiotherapy (nCRT) in patients with locally advanced rectal cancer (LARC).

Methods

We retrospectively enrolled 226 patients with LARC [allocated in a 7:3 ratio to a training (n = 158) or validation (n = 68) cohort] who received nCRT before radical surgery. Backward stepwise logistic regression was performed to identify clinical and blood parameters associated with achieving pCR. Models based on clinical parameters (CP), blood parameters (BP), and clinical-blood parameters (CBP) were constructed for comparison with previously reported Tan's model. The performance of the four models was evaluated by receiver operating characteristic (ROC) curve analysis, calibration, and decision curve analysis (DCA) in both cohorts. A dynamic nomogram was constructed for the presentation of the best model.

Results

The CP and BP models based on multivariate logistic regression analysis showed that interval, Grade, CEA and fibrinogen-albumin ratio index (FARI), sodium-to-globulin ratio (SGR) were the independent clinical and blood predictors for achieving pCR, respectively. The area under the ROC curve of the CBP model achieved a score of 0.818 and 0.752 in both cohorts, better than CP (0.762 and 0.589), BP (0.695 and 0.718), Tan (0.738 and 0.552). CBP also showed better calibration and DCA than other models in both cohorts. Moreover, CBP revealed significant improvement compared with other models in training cohort (P < 0.05), and CBP showed significant improvement compared with CP and Tan's model in validation cohort (P < 0.05).

Conclusion

We demonstrated that CBP predicting model have potential in predicting pCR to nCRT in patient with LARC.