This is a very well-designed and well-written study investigating the role of NSAIDs on HNSCC prognosis and the underlying mechanisms of the identified survival benefit among PIK3CA mutated or amplified cases. It is a high impact paper identifying a potentially causative relationship between NSAIDs and survival and also describes the underlying molecular mechanism of this association. Provided that these findings are validated prospectively in RCTs, NSAIDs could change the therapeutic landscape of HNSCC, providing a valuable, low-cost addition to the available options.

Major points

1. The authors use both p16 IHC and HPV ISH for HPV classification. This is very useful to avoid HPV negative p16 positive cases. However, they end up including some cases based exclusively on p16 expression. p16 based assays have are reportedly misclassifying 5-20% of OPSCC cases (DOI:https://doi.org/10.1093/annonc/mdy110). Given the impact of HPV status on OPSCC prognosis, case misclassification could significantly affect the outcome and thus, should be be avoided

2. The authors report an adjustment set that does not include T-stage/ TNM stage, and treatment parameters (e.g. surgery only vs surgery + aRT, surgery +CRT). These would be important variables to be included in the adjustment set.

3. Chemotherapy treatment presents some heterogeneity ( Supplementary Table 2 e.g. Cisplatin recipients: 21% and Cetuximab 6%). Given the substantially different underlying mechanisms of these treatments, the inclusion of cetuximab could have been avoided in order to ensure study homogeneity (since it is hard to compare the 2 groups, given the low numbers of cetuximab and other non-cisplatin therapy recipients).

4. Patients reportedly received several non-steroid anti-inflammatory drugs (Supplementary Table 3). Specifically, they did receive aspirin (non-selective, irreversible COX inhibitor), Ibuprofen (non-selective, reversible COX inhibitor) or Celecoxib (reversible COX-2 selective inhibitor). Also, the authors used Celecoxib, Indomethacin (non-selective, reversible COX inhibitor), and Sulindac (non-selective, reversible COX inhibitor) in-vitro. It is not clear how this variability of NSAIDs could affect the outcomes, so it is suggested that the authors either compare the different mechanisms’ outcomes (reversible vs irreversible, specific vs non-specific) or stick with one drastic substance.

5. The authors mention that the cause of NSAID use was not related to the patients HNSCC diagnosis, but they do not specify the exact reason(s) for drug administration. This would be a valuable addition to the paper.

6. The authors are encouraged to comment on the paper by Bosetti et al., 2019 ( DOI: 10.1016/j.annonc.2020.02.012). In this meta-analysis, the authors identify a significant survival benefit for aspirin recipients in most aerodigestive tract carcinomas, but not in HNSCC. While they do not stratify for PIK3CA status (probably due to data unavailability) the study has the advantage of focusing on only 1 NSAID and also including multiple HNSCC studies (n=10). It would be important for the authors to discuss their findings in the context of this meta-analysis, which is to my knowledge, the largest and most comprehensive one for this topic. Why in their opinion did this study failed to identify a survival benefit?

Minor points

1. The authors mention that current methodologies are unable to assess PGE2 expression in FFPE specimens. I am not sure that this is entirely true, because RNA sequencing methods actually do provide info on gene expression and thus can be of some help in comparing quantitatively PGE2 expression (e.g. DOI: 10.1007/978-1-4939-0700-7_5).