Pancreatic Ductal Adenocarcinoma: A New TNM Staging System Is Needed!
We thank Drs Crippa, Partelli, and Falconi for their interest in our article and we entirely share their opinion that a new TNM staging is (or was) needed.1 Indeed our final conclusion was that N-categories based on the number of positive lymph nodes (PLN) should be included in a revised TNM system after validation of our data in independent cohorts.1 Meanwhile, our findings were confirmed by 2 independent cohort studies.2,3
With the relevance of the site of PLN, Crippa et al address an important topic. As we specified in our article, the exact location of regional PLN was not routinely reported and, therefore, not available for our study.1 As also stated, interaortocaval lymph nodes, including station 16, were considered extraregional, not included in standard lymphadenectomy, and only removed if suspicious. Positive interaortovacal/paraaortic lymph nodes were identified in n = 25 patients. Although these patients were excluded from the main analysis due to pM1 stage, their median survival was 13.6 months and poorer than in patients with at least 8 regional PLN (18.3 months), consistent with a more advanced lymphatic (or systemic) spread.1 With respect to the relevance of the site of regional PLN, we would like to refer to Malleo et al.3 They analyzed the site of PLN in a cohort of 255 patients and found by multivariable analyses that not only the number of PLN but also the number of PLN stations and the site of PLN were associated with survival. In their study, PLN at station 14a-b were associated with poorer survival.3 Although the site of PLN may allow to predict prognosis with an even higher resolution that the number of PLN alone, it would add too much complexity to be included in a revised TNM staging system. The next logical question would then be how the frequencies and prognostic impact of PLN at different sites vary with tumor location within the pancreas. While the cohort studies mentioned above were performed by specialized centers and are based on high numbers of examined lymph nodes, a revised TNM staging system will have to be applicable on a broad level. In this regard, the validation of PLN-based categories by population-based studies and the clarification of the necessary number of examined lymph nodes to determine these categories are more relevant questions that remain to be addressed. In an analysis based on the Surveillance, Epidemiology, and End Results database from 2004 to 2012 with inclusion of cases with at least 12 examined lymph nodes, we could confirm the possibility of PLN-based categories for refined staging and identify <=2 versus >=3 PLN as the optimal cutoff.4
The next important point raised by Crippa et al was the necessity to revise the T-classification, based on the observations that (i) with the current classification the vast majority of 95.5% of resected pancreatic cancers are staged pT3, resulting in a poor relevance of pT-stages, and (ii) portal vein invasion was shown to be associated with decreased survival, a fact not considered in the current staging.5,6 We certainly agree that the current pT-classification results in a very imbalanced distribution that hampers the value of this variable for prognostication. In our study, 95.3% of all tumors and 97.7% of N-positive tumors were pT3; pT-stage was associated with survival by univariable analysis, but this was not confirmed by multivariable analysis.1 However, we are not sure if there is sufficient evidence to use portal vein invasion to define pT-stage. Many venous resections are performed for inflammatory adhesions rather than for true invasion.