Evaluation of Neoadjuvant Chemotherapy in Patients with Resectable Gastric Adenocarcinoma in Central Norway 2007 – 2016
Abstract
This thesis presents three retrospective, population-based, observational studies on patients diagnosed with upfront resectable gastric adenocarcinoma and receiving neoadjuvant chemotherapy (NAC) in Central Norway during the period from 2007-2016. NAC has since 2007 been given as a onesize- fits-all treatment to patients in a curative setting with disease stage Ib (T1N1)- III (according to the 7th UICC edition), age below 75 years, and good health (WHO 0-1). However, increasing evidence suggests a differential response to oncological treatment according to variables such as tumour location, Lauren type, microsatellite instability etc. No consensus exists on how to perform a response evaluation following NAC, neither what a response to neoadjuvant treatment is nor the impact of a response.
Our first study focused on the radiological response to NAC and the consequences of a restaging CT for patients with tumours of true gastric and Siewert cardia types II/III location. We explored how standardised CT scans performed when used to evaluate a strict radiological response following NAC in terms of UICC-downstaging, that is, the CT-stage at baseline was compared to the CT-stage following NAC. Further, we correlated the radiological stage following NAC (yrTNM) to the final pathological stage (ypTNM) following resection to assess CT-restaging accuracy after NAC. Our main finding was that a radiological UICC-restaging following NAC (yrTNM) was highly unreliable compared to the final pathological stage (ypTNM) and that a strict CT downstaging, as a consequence, was an inadequate tool for assessment of the response to NAC. No patient developed deep organ metastases during NAC, and the CT scan following NAC had a limited ability to detect peritoneal carcinomatosis. The restaging CT did not add to decision-making concerning surgery. We concluded that the use of routine CT scans following NAC in gastric cancer are not warranted.
The second study is a logical continuation based on the findings in Study 1. We wanted to investigate alternative methods of response evaluation using the same patient cohort. We conducted two methods of response evaluation, a radiological downsizing using the Response Evaluation Criteria in Solid Tumours (RECIST) and an alternative method of downstaging by comparing the radiological stage at diagnosis (rTNM) to the pathological stage following chemotherapy (ypTNM). We aimed to disclose how RECIST performed when applied to a Western cohort of resectable gastric cancer, since the use of RECIST is not validated for neither gastric cancer nor the setting of neoadjuvant treatment but is still frequently used in both contexts. RECIST failed to assign patients to subsets with different long-term survival rates, whereas the composite method of assessing the TNM response mode did achieve this objective. In a multivariable ordinal regression model, the Lauren type and the preoperative tumour location were significant determinants of the response mode. We concluded that comparison of the CT stage at diagnosis (rTNM) to the pathological stage following NAC (ypTNM) appears to be a useful method in evaluating the response to NAC and can be a tool to serve the reallife situation. Patients judged to respond to NAC (downstaged) seemed to have a true survival benefit.
The third study set out to determine whether ypTNM survival curves follow corresponding stagematched survival curves in chemo naïve patients (pTNM). That is, does the pathological stage after neoadjuvant chemotherapy (ypTNM) predict long-term survival rates as seen in upfront radically resected patients? The final ypTNM stage could theoretically merely represent a “cosmetic” downstaging, not necessarily reflecting the range of possibly non-eradicated initial disease spread, and its capabilities as a prognostic tool needs to be proven. Further, we also wanted to disclose if there was any interaction between the TNM response mode and the stage-specific long-term survival rates. We concluded that the pathological TNM staging system seems to adequately predict prognosis for radically resected patients with gastric adenocarcinoma, even after receiving NAC. Long-term survival rates for a particular ypTNM stage followed the corresponding curve for stage-matched chemo naive patients (pTNM). We found no interaction between the response mode and the stage-specific longterm survival rates. The final pathological disease stage following NAC determined survival regardless of whether the patient was assigned that disease stage through responding to NAC or remained stable/upstaged. E.g., survival rates for the stage IIIA disease were equal regardless of whether downstaged from an initial stage IIIC disease or being an upstaged stage IIA disease.