Interobserver variability in the H&E-based assessment of tumor budding in pT3/4 colon cancer: does it affect the prognostic relevance?

Abstract

Tumor budding is a mostly accepted adverse prognostic factor in colorectal carcinoma. It is on the cusp of a widespread use after agreement was reached recently on uniform assessment criteria. We investigated whether the interobserver variability has a direct influence on the prognostic relevance in pT3/4 colon cancer in the background of different levels of experience of the investigators. In total, six investigators with different levels of experience evaluated tumor budding on H&E slides in 244 cases with primary diagnosed (2002-2011) colon carcinoma (pT3/4, N+/-, M0). High-grade tumor budding/budding grade 3 (defined as majority assessment among the investigators) was significantly associated with an adverse outcome (overall survival p = 0.03, cancer-specific survival p = 0.08) and the occurrence of distant metastasis (p = 0.009). However, a detailed analysis of the rating results of the individual investigators revealed that only ratings of one investigator (advanced resident) were associated with an adverse outcome (p = 0.01 cancer-specific survival, overall survival p = 0.09, distant metastasis p = 0.002). The results of another investigator (consultant) were significantly associated with distant metastasis (p = 0.007). The kappa values among the investigators have a range between 0.077 and 0.357 (median 0.166). Total agreement of all investigators existed in 109 cases (44.7%). Our results demonstrate that the evaluation of tumor budding on H&E slides in pT3/4 colon cancer goes along with a considerable interobserver variability among investigators of different levels of experience. Furthermore, our results reveal that these findings directly influence the prognostic value.

Level of interobserver variability estimation as a valuable tool: assessment of tumour budding in colon cancer

“…In this context it is worth emphasising that most investigators of this study were residents. Moreover, the training period was limited . Following the idea that the question of ‘how difficult is the assessment?’ is a risk stratification, there has to be a consequence for D3/Li3 classified cases.…”

“…Following the idea that the question of ‘how difficult is the assessment?’ is a risk stratification, there has to be a consequence for D3/Li3 classified cases. The mandatory dialogue and the corporate assessment, together with a colleague, might provide a better assessment quality and cytokeratin staining might be helpful in these cases. However, it has to be shown that the two techniques in fact have a measurably reductive effect on the interobserver variability in these cases.…”

“…Furthermore, the investigators had to estimate the interobserver variability/level of interobserver variability (LOIV) [1 = low/lower than average, 2 = average, 3 = high/above average (Li1, Li2 and Li3)]. Detailed methods of the study have been published previously . In brief, six investigators with different levels of experience (two specialists, four residents) evaluated tumour budding in an approximately everyday routine setting according to the ITBCC method (exception: only one exemplary slide per case).…”

“…The International Tumour Budding Consensus Conference (ITBCC) recommended including tumour budding into future colorectal cancer guidelines . The obstacle of different assessment criteria has been overcome, but the significance of reliability in daily practice has to be clarified further . The method mainly investigated in the context of reduction of the interobserver variability in tumour budding is the use of immunohistochemistry (anti‐cytokeratin) .…”

“…4 The obstacle of different assessment criteria has been overcome, 4 but the significance of reliability in daily practice has to be clarified further. 5 The method mainly investigated in the context of reduction of the interobserver variability in tumour budding is the use of immunohistochemistry (anti-cytokeratin). [6][7][8][9] The ITBCC criteria recommend the evaluation of tumour budding on haematoxylin and eosin (H&E).…”