Background and objectives
In the WHO 2010 classification, resection specimens of pancreatic neuroendocrine tumors (pNETs) are graded using the Ki67-labeling index (Ki67-LI) (G1 : Ki67-LI < 2% ; G2 : 3-20%, G3 : > 20%). These past few years, endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) has become an important diagnostic tool of pNETs by collecting cytological samples. Although many studies have considered the diagnostic accuracy of EUS-FNA, only few have dealt with grading of pNETs in EUS-FNA. This study is an extension of a previously published paper from our team [1] that assessed prognostic value of Ki67-LI on EUS-FNA in 46 pNETs (of which 33 were surgically resected). The aim of this study is to compare (grades and raw values of) Ki67-LI on cytological analysis (FNA) with the ones obtained on surgical specimens. Analysis regarding influence of tumor size and number of counted cells in FNA grading will secondly be addressed, along with overall survival (OS) and progression free survival (PFS) estimates of all patients based on cytological grade.
Methods
Between 1996 and 2013, 102 pNETs from 101 patients (57 required surgery) were retrospectively included in this multicentered study. All of them underwent EUS-FNA (22 or 25-gauge needle) at the time of diagnosis. Cytological Ki67-LI was evaluated on FNA material of the 102 tumors (200 cell count). In a subgroup of 29 FNA specimens, more than 2000 cells were counted (14 patients underwent surgery). For patients who underwent surgery, Ki67-LI of resected tumor was assessed (more than 2000 counted cells) and compared with Ki67-LI of the corresponding FNA specimen. All patients were followed-up until June 2016.
Results
Cytological grade was consistent with histological grade in 39/57 cases hence a concordance rate of 68.4% when using a 3% cut-off between G1 and G2 tumors (72% if the cut-off is 5%). Agreement between FNA grade and surgical specimen was significant using a 3% (k=0.434, p<0.001) or a 5% cut-off (k=0.354, p<0.001). Concerning Ki67-LI absolute values, correlation remained significant (r=0.443, p=0.001) and raised when more tumor cells were counted (r=0.574, p<0.001 when FNA samples with less than 200 cells were excluded; r=0.824, p<0.001 when including only cases with more than 2000 counted cells). Mean tumor size was significantly smaller when cytological and histological grading was consistent (26 vs. 35mm, p=0.023, 5% cut-off). Thirty-eight of 101 patients died during a median follow-up of 70.5 months. The median overall survival (OS) of the entire population is 235.30 months. OS is significantly different between tumor grades based on cytological Ki67-LI (log rank test, p<0.001) with a 3% cut-off (G1 235.30 months, G2 68.68 months and G3 10.95 months) and a 5% cut-off (G1 235.30 months, G2 36.35 months and G3 10.95 months; HR vs. G1 : 3.78 and 12.55). The median progression free survival (PFS) is significantly greater (log rank test, p<0.001) for patients with a G1 tumor than for those with a G2 (39.80 months) or a G3 (10.07 months) tumor (HR vs. G1 : 2.61 and 14.70).
Conclusions
The current results indicate that pNETs cytological grading is accurate when tumor size is < 2 cm and more tumor cells are counted on FNA. Discrepancies are seen among G2 tumors that are often considered G1 on FNA material due to tumor heterogeneity. Nevertheless EUS-FNA is a valuable tool to distinguish patients with a good (G1 tumors) or a poor (G3 tumors) prognosis, in terms of both OS and PFS. Patients with G2 and G3 pNET have respectively 3.78 and 12.55 times higher risk of death than G1 patients and are 2.6 and 14.7 times more at risk of disease progression. Our results show that EUS-FNA is helpful to clinicians by providing important prognostic information leading to adequate therapeutic decisions.
[1] Weynand B., Borbath I., Bernard V., et al. Pancreatic neuroendocrine tumour grading on endoscopic ultrasound-guided fine needle aspiration: high reproducibility and inter-observer agreement of the Ki-67 labelling index. Cytopathology 2014, 25:389-95.