Predicting pathological response to neoadjuvant therapy in esophageal cancer

To date, various neoadjuvant chemotherapy, chemoradiotherapy, immunotherapy regimens and combinations thereof, have been proposed for locoregional esophageal cancer to increase pathologic complete response (ypCR) rates as pCR has been regarded to be a surrogate to higher overall survival and lower recurrence rates (1-3). However, oncologists and thoracic surgeons alike struggle to identify the cohort of patients who may have CR on their final pathology. CR allows these patients to be committed to a path of watchful waiting and then salvage esophagectomy in case they have a recurrence, where the results of resection are not significantly inferior to conventional planned esophagectomy performed 6–8 weeks after completion of neoadjuvant therapy (4). In fact, more and more surgeons and patients rather opt to have clinical CR (cCR) in order to avoid a morbid procedure such as an esophagectomy. However, correlating ycCR to ypCR after neoadjuvant chemoradiation for locoregional, i.e., resectable, esophageal cancer has been a topic of study for over a decade now (5-7).

Unfortunately, despite all the clinical prediction tools including computed tomography (CT) scans, positron emission tomography (PET) scans, endoscopies, esophageal ultrasound (clinical T, clinical N, tumor length, tumor burden), and biopsies including tumor regression grade, scientists have not been able to reliably predict ypCR with 100% accuracy (6,8-11). Multiple proteomic, hematologic and immunologic tools have also been integrated into various prediction models and nomograms to predict ypCR and overcome this challenge (12-15). Having said that, perhaps it is time for surgeons and oncologists to start integrating all these parameters into nomograms and then implementing these nomograms into practice as a way to see if we can:

• Accurately identify complete or incomplete responders especially in those patients who elected to undergo surgery, even when the nomogram predicted ypCR; and compare them with those who were deemed non-ypCR, but surprisingly turned out to be ypCR. This would help us analyze what predictive features were missed or “under-called” and how could we refine prediction of ypCR in the future.
• Stratify patients who can be monitored vs. proceed with resection. And then longitudinal studies ought to be conducted to validate whether the decision to wait and watch was the correct path to take. Time from decision of deferring surgery to recurrence or resurgence of metastatic disease would determine if the decision to wait and watch was the right one. Such a study would also help determine whether ypCR is truly a surrogate of becoming truly “cancer-free”.

Very recently, Fan et al. published a study where they constructed a nomogram as their mathematical model of choice to correlate medical imaging with the ability to predict ypCR (16). Using artificial intelligence (AI), they extracted quantitative features from pre-treatment and post-treatment scan of 186 patients with cT1–4, N0–3 esophageal squamous cell carcinoma (eSCC). Then with quantitative analysis and specific algorithms (PyRadiomics software), they calculated a delta-radiomics feature (DRF) and extrapolated delta-radiomics signature (DRS). This difference is considered to be reflective of underlying pathophysiology as a response to neoadjuvant therapy. Further, they cross-validated this machine learning model by re-sampling and running the same algorithm on a limited sample of 46 patients from the same institution. Using least absolute shrinkage and selection operator (LASSO) algorithm and α-binormal unequal-variance model, the authors demonstrated how radiomics correlated with binary classification of ypCR and non-ypCR. However, there are multiple challenges as rightly pointed out by this study including some gaps that do not allow translation of the study results into practice:

• Due to poor quality of images (despite this being a single institution study), Fan et al. were unable to include 5.9% of the 304 patients in their final analysis who initially met the study criteria. This would be one of the limitations in the application of such a methodology to predict ypCR.
• The study also had several inappropriate and inconsistent groupings of cT and cN staging variables, degree of differentiation, as well as Eastern Cooperative Oncology Group (ECOG) performance status [refer to Tab. 1 in Fan et al. (16)]. The study included 43 cT1–2 patients (albeit the N-paired status is not clear) in the training dataset analysis. Classically, the treatment of cT1 lesions and even cT2 lesions (presumably node negative) should have been an endoscopic resection or an esophagectomy. Neoadjuvant therapy in this cohort of patients is not standard of care per NCCN guidelines. However, if the lesion was cT2, node-positive, >3 cm in size, with high-risk features such as lymphovascular invasion or positive margins and/or poorly differentiated, one could argue the use of neoadjuvant therapy.
• The authors also grouped low and middle-high differentiation together, when ideally they should have left all 3 categories separate. The authors failed to define differentiation nomenclature, as conventionally scientists classify esophageal cancer patients as poor, moderate, or well-differentiated. In fact, even when reviewing the nomogram [Figure 3 in Fan et al. (16)], the classification system is not clear as low-differentiated tumors were assigned a lower score thus suggesting that low-differentiated tumors are synonymous with poorly-differentiated tumors thus giving them a lower chance of achieving ypCR after neoadjuvant therapy.
• Lastly, the authors extracted 93 radiomic features from regions of interest from 3D slicer software using Python package, calculated a DRF and then translated “4 of these features” into DRS, however, the authors fail to explain which features were selected in univariate analysis, before LASSO, and then used for logistic regression for multivariable analysis. Such details are necessary to understand the development of the prediction model and construction of nomogram as well as for replication/validation of results. Finally, the authors used a term “radiomics score” in their nomogram but never clearly defined exactly what this term referred to.

Despite all the shortfalls in the study, Fan et al. have introduced a novel approach of utilizing delta-radiomics and applying that knowledge to prediction of ypCR after neoadjuvant therapy in eSCC (16). Over time, it will be important to study the clinical application of this specific model to populations outside of China where clinical factors and management, as well as environmental and genetic factors are likely to differ—thus modulating the probability of pCR. The benefit of this methodology lies in the radiomics—the ability to extract quantitative features, without relying solely on qualitative examination from a radiologist.

We, the authors, hereby strongly propose the adoption of routine utilization of a nomogram-like prediction tool. A combination of clinical factors, radiomics, immunochemical factors (12), and/or proteomics (13) may indeed need to be incorporated into a comprehensive nomogram (17). As the landscape of oncology evolves with the emergence of new biologics and immunotherapies as well as new radiation protocols along with dose de-escalation and hypofractionation (18-20), such a nomogram may have to be revised.

Acknowledgments

None.

Footnote

Provenance and Peer Review: This article was commissioned by the editorial office, Annals of Esophagus. The article has undergone external peer review.

Peer Review File: Available at https://aoe.amegroups.com/article/view/10.21037/aoe-24-33/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://aoe.amegroups.com/article/view/10.21037/aoe-24-33/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Verschoor YL, van de Haar J, van den Berg JG, et al. Neoadjuvant atezolizumab plus chemotherapy in gastric and gastroesophageal junction adenocarcinoma: the phase 2 PANDA trial. Nat Med 2024;30:519-30. [Crossref] [PubMed]
2. Zhou Y, He W, Guo P, et al. Development and Validation of a Recurrence-Free Survival Prediction Model for Locally Advanced Esophageal Squamous Cell Carcinoma with Neoadjuvant Chemoradiotherapy. Ann Surg Oncol 2024;31:178-91. [Crossref] [PubMed]
3. Khaitan PG, Ponnusamy S. Comparison of the treatment outcomes of different neoadjuvant chemoradiotherapy regimens for resectable locally advanced esophageal cancer. J Gastrointest Surg 2024;28:2148-9. [Crossref] [PubMed]
4. Pan J, Liu Z, Yang Y, et al. Salvage Esophagectomy After Definitive Chemoradiotherapy for Squamous Cell Esophageal Cancer: A Propensity Score Matching Study in a High-Volume Center. World J Surg 2023;47:2003-12. [Crossref] [PubMed]
5. Lin SH, Wang J, Allen PK, et al. A nomogram that predicts pathologic complete response to neoadjuvant chemoradiation also predicts survival outcomes after definitive chemoradiation for esophageal cancer. J Gastrointest Oncol 2015;6:45-52. [PubMed]
6. Khaitan PG, Holliday T, Carroll A, et al. Can Clinical Response Predict Pathologic Response Following Neoadjuvant Chemoradiation for Esophageal Cancer? J Gastrointest Surg 2022;26:1345-51. [Crossref] [PubMed]
7. Liu G, Chen T, Zhang X, et al. Nomogram for predicting pathologic complete response to neoadjuvant chemoradiotherapy in patients with esophageal squamous cell carcinoma. Cancer Med 2024;13:e7075. [Crossref] [PubMed]
8. Uberoi GS, Uberoi AS, Bhutani MS. Endoscopic and Imaging Predictors of Complete Pathologic Response After Chemoradiation for Esophageal Cancer. Curr Gastroenterol Rep 2017;19:57. [Crossref] [PubMed]
9. Ajani JA, Correa AM, Hofstetter WL, et al. Clinical parameters model for predicting pathologic complete response following preoperative chemoradiation in patients with esophageal cancer. Ann Oncol 2012;23:2638-42. [Crossref] [PubMed]
10. Gaur P, Sepesi B, Hofstetter WL, et al. A clinical nomogram predicting pathologic lymph node involvement in esophageal cancer patients. Ann Surg 2010;252:611-7. [Crossref] [PubMed]
11. Gaur P, Sepesi B, Hofstetter WL, et al. Endoscopic esophageal tumor length: a prognostic factor for patients with esophageal cancer. Cancer 2011;117:63-9. [Crossref] [PubMed]
12. Feng J, Wang L, Yang X, et al. Pathologic Complete Response Prediction to Neoadjuvant Immunotherapy Combined with Chemotherapy in Resectable Locally Advanced Esophageal Squamous Cell Carcinoma: Real-World Evidence from Integrative Inflammatory and Nutritional Scores. J Inflamm Res 2022;15:3783-96. [Crossref] [PubMed]
13. Han D, Han Y, Guo W, et al. High-dimensional single-cell proteomics analysis of esophageal squamous cell carcinoma reveals dynamic alterations of the tumor immune microenvironment after neoadjuvant therapy. J Immunother Cancer 2023;11:e007847. [Crossref] [PubMed]
14. Wu Y, Chen J, Zhao L, et al. Prediction of Pathologic Response to Neoadjuvant Chemoradiotherapy in Patients with Esophageal Squamous Cell Carcinoma Incorporating Hematological Biomarkers. Cancer Res Treat 2021;53:172-83. [Crossref] [PubMed]
15. Li ZM, Liu W, Chen XL, et al. Construction and validation of classification models for predicting the response to concurrent chemo-radiotherapy of patients with esophageal squamous cell carcinoma based on multi-omics data. Clin Res Hepatol Gastroenterol 2024;48:102318. [Crossref] [PubMed]
16. Fan L, Yang Z, Chang M, et al. CT-based delta-radiomics nomogram to predict pathological complete response after neoadjuvant chemoradiotherapy in esophageal squamous cell carcinoma patients. J Transl Med 2024;22:579. [Crossref] [PubMed]
17. Borggreve AS, Mook S, Verheij M, et al. Preoperative image-guided identification of response to neoadjuvant chemoradiotherapy in esophageal cancer (PRIDE): a multicenter observational study. BMC Cancer 2018;18:1006. [Crossref] [PubMed]
18. Shitara K, Rha SY, Wyrwicz LS, et al. Neoadjuvant and adjuvant pembrolizumab plus chemotherapy in locally advanced gastric or gastro-oesophageal cancer (KEYNOTE-585): an interim analysis of the multicentre, double-blind, randomised phase 3 study. Lancet Oncol 2024;25:212-24. [Crossref] [PubMed]
19. Park H, Yang K, Noh JM, et al. Comparison of the treatment outcomes of different neoadjuvant chemoradiotherapy regimens for resectable locally advanced esophageal cancer. J Gastrointest Surg 2024;28:1745-51. [Crossref] [PubMed]
20. Mukherji R, Alqahtani A, Yin C, et al. Perioperative therapy landscape for locally advanced, resectable esophageal cancer: an updated literature review. J Thorac Dis 2023;15:3466-87. [Crossref] [PubMed]