The canary in the coal mine: the growth of patient-derived tumorgrafts in mice predicts clinical recurrence after surgical resection of pancreatic ductal adenocarcinoma

Abstract

Background: Recurrence after resection of pancreatic ductal adenocarcinoma (PDAC) is common, thus postoperative surveillance is critical for detection and treatment of recurrent disease. The development of biologically based techniques for early recurrence detection may enable more timely and effective treatment of such recurrences.

Methods: Tumor fragments derived from patients who underwent potentially curative resection of PDAC were heterotopically implanted into NOD/SCID mice. Engraftment success rates and growth parameters were matched to clinicopathologic data, preoperative treatment status, and oncologic outcomes to correlate disease-free survival (DFS) and overall survival.

Results: Seventy patients consented to participate with 56 (80 %) developing a mouse PDAC tumorgraft. Patients with successful engraftment had a shorter median DFS compared with patients whose tumorgrafts failed to engraft (9.8 vs. 40.9 mo, respectively; p < 0.01). Fifty patients received preoperative therapy with 36 (72 %) successful tumorgrafts from this cohort. On multivariate analysis, lymph node metastasis (hazard ratio [HR] 3, 95 % CI 1.4-6.7, p < 0.01) and successful engraftment (HR 5.8, 95 % CI 2-16.9, p < 0.01) were predictive of a shorter DFS in the preoperative therapy cohort. In patients who recurred, tumorgraft formation was identified at a median of 134.5 days before standard methods of radiographic recurrence detection (p < 0.01).

Conclusions: Patient-derived tumorgrafts from resected PDAC may potentially predict recurrence months before currently available surveillance modalities. This lead-time advantage may allow for earlier implementation or changes in therapy as successful engraftment, particularly in those having undergone preoperative therapy, may indicate a more biologically aggressive disease.

FIG. 1

Flow diagram of the cohort based on the treatment strategy of surgery first (a) versus preoperative therapy followed by surgery (b). Each arm illustrates the number of primary tumor implantations attempted and the number of subsequent tumorgrafts established. Total number patients in each group (n), number of recurrence events with median DFS, and number of deaths with median OS are indicated for each patient group

FIG. 2

Kaplan-Meier analysis comparing the entire cohort of patients who underwent potentially curative resection of PDAC from February 2008 to July 2012, based on attempted tumorgraft implantation, did not show a statistical difference in DFS (a) or OS (b) between the two groups. Patients who underwent preoperative therapy and had successful tumorgraft growth had a significantly shorter DFS (c) and OS (d) compared with patients who underwent preoperative therapy yet did not have successful growth of their tumorgraft

FIG. 3

The timing of each patient’s postoperative PDAC recurrence was matched to the timing of their corresponding individual tumorgraft growth. This paired analysis demonstrated that tumorgraft formation occurred a median of 134.5 days prior to the detection of a clinical recurrence (p \ 0.01)