Correlation between pancreatic tumor size as measured on 3D CT scan versus pathologic specimen: Impact on radiation treatment volume

276

Background: Definition of the target volume for irradiation of pancreatic cancer (PCA) must balance coverage of micrometastatic disease with toxicity. To determine differences between radiographically defined tumors and true pathologic tumor specimens, we correlated the maximum tumor diameter (TD) of preoperatively imaged tumors with resected tumor specimens.

Methods: With IRB approval, a retrospective chart review of patients who underwent resection of PCA between 2006 and 2010 was conducted. 73 patients were identified with preoperative CT imaging and pathologic analysis of tumors. 70 of 73 patients had a preoperative 3D CT performed. The TD as measured by a radiologist (EF) on contrast CT and 3D CT reconstruction was compared with that measured by pathological analysis of the resected specimen.

Results: 70 patients underwent resection with preoperative CT imaging; 14.1% of these patients had CT performed >6 weeks prior to surgery. The mean (SD) pathologic maximum TD was 31.3 mm (11.3) with range 3 mm to 60 mm. Whereas TD was underestimated by 1.9 mm (1.7 SE) with CT relative to pathologic analysis, this difference was not statistically significant (paired t-test, p=0.27) with a correlation coefficient of 0.265. 3D CT imaging had a smaller mean difference with a mean 3D CT diameter 0.4 mm (1.76 SE) larger than the pathologic specimen (p=0.82) with correlation coefficient 0.222. However, the max TD on 3D CT imaging was on average 2.3 mm larger than on CT (p=0.016) with correlation coefficient 0.798. Of patients with R0 resections (N=48), CT underestimated path size by 3.1 mm (p=0.020), whereas 3D CT was slightly larger (0.1 mm, p=0.949). For R1 resections (n=22), both CT and 3D CT overestimated size (0.8 mm and 1.1 mm, respectively, p>0.5).

Conclusions: PCA TD is generally underestimated on CT imaging, yet better approximated with 3D CT. Improved correlation was seen between CT and pathologic specimens following R0 resection. Alternatively, R1 resection specimens were slightly overestimated by CT/3D CT imaging. As a result, clinical target volumes should be expanded accordingly during radiotherapy planning to properly account for these discrepancies in the gross tumor.

No significant financial relationships to disclose.