Patients’ Characteristics

Mean patient age was 65 ± 10 (range 44–76) years (Table 1). Most patients were male (13 of 16; 81%). Median Ki67-score was 80 (range 45–90)%. Most patients were treatment naïve (12 of 16; 75%) at the time of CXCR4 PET/CT. Mean time between initial diagnosis and CXCR4 PET was 9 ± 9 (range 2–23) months for those patients (4 of 16) who were pretreated. Median time interval between CXCR4 and FDG PET (15 of 16 patients; 94%) was 6 days (range 1–53 days). Assessment of rPFS was possible in 14 of 16 patients (88%), because follow-up imaging was not available in 2 of 16 patients (12%).

Table 1 Patients’ characteristicsCXCR4 PET-Derived Volumetric Parameters Are Prognostic for OS and rPFS

Mean parameters for TV, TLU, SUVmax, SUVpeak, and SUVmean were 83.0 ± 112.1 ml (range 0–476.6 ml), 391.3 ± 443.7 (range 0–2215.5), 9.6 ± 5.0 (range 0–18.5), 7.3 ± 4.3 (range 0–18.5), and 4.5 ± 2.3 (range 0–8.2), respectively. Of note, 2 of 16 patients (13%) did not show any pathological uptake on CXCR4 PET, despite liver and lymph node metastases seen on corresponding FDG PET/CT. Disease progression occurred in all patients. Thirteen of 16 patients (81%) died during follow-up. Median OS and rPFS of the whole patient cohort was 7.5 (n = 16) and 7 (n = 14) months, respectively. A significant correlation between TV and TLU was found for OS (TV: hazard ratio (HR) 1.007, 95% confidence interval (CI) 1.000–1.014, p = 0.0309; TLU: HR 1.002, 95% CI 1.000–1.003, p = 0.0350) and rPFS (TV: HR 1.010, 95% CI 1.002–1.021; p = 0.0275; TLU: HR 1.002, 95% CI 1.000–1.004, p = 0.0329), respectively. No significant correlation with neither OS nor rPFS was found for non-volumetric parameters (p > 0.4, respectively).

FDG PET-Derived Parameters Are Not Prognostic for OS and rPFS

Mean parameters for TV, TLU, SUVmax, SUVpeak, and SUVmean were 243.7 ± 229.0 ml (range 1.8–785.9 ml), 2422.4 ± 2510.7 (range 8.7–7758.4), 18.6 ± 10.7 (range 6.5–43.3), 14.0 ± 7.9 (range 4.5–29.7), and 9.0 ± 4.8 (range 3.5–20.8), respectively. The two patients with negative CXCR4 PET showed intense uptake on FDG PET (TV: 18.4 ml, TLU: 164.1, SUVmax: 14.7, SUVpeak: 10.3, SUVmean: 8.9 and TV: 433.9 ml, TLU: 7341.6, SUVmax: 37.1, SUVpeak: 29.7, SUVmean: 16.9). All PET-based parameter were significantly higher FDG PET compared to CXCR4 PET (p-values: TV: 0.0467, TLU: 0.0097, SUVmax: 0.0075, SUVpeak: 0.0082, SUVmean: 0.0046; Fig. 1). Median OS and rPFS of those patients who underwent an additional FDG PET/CT was 6 (n = 15) and 5.5 (n = 13) months, respectively. None of the volumetric or non-volumetric parameters were prognostic for OS or rPFS (p > 0.06). Table 2 summarizes the outcome data of CXCR4 and FDG PET-derived parameters. Figure 2 depicts different patterns of tracer uptake on CXCR4 and FDG PET.

Fig. 1

Paired t-test of volumetric parameters (A) and standardized uptake values (SUV; B) of CXCR4 (blue bars) and FDG (orange bars) PET derived parameters. The scale bars indicate the 5th to 95th confidence interval. Abbreviations: TV, tumor volume; TLU, total-lesion uptake (= TV * SUVmean)

Table 2 Univariable cox regression of OS and rPFS of CXCR4 and FDG PET derived parametersFig. 2

Comparison of PET-based biomarker of corresponding maximum intensity projections of CXCR4 (upper row) and FDG (lower row) PET. Tumor is delineated using Fiji [19] and the Beth Israel Plugin [20] and marked in blue. SUV window ranging from 0 to 5. A, B A 75-year-old patient with pretreated gastric neuroendocrine carcinoma (NEC) and a negative CXCR4 PET scan with missing tracer uptake of the liver metastases. C, D A 55-year old patient with initial diagnosis of an esophageal NEC and metastases in lymph nodes, liver, bones, and the lungs. E, F A 68-year-old patient with initial diagnosis of a pancreatic NEC and local lymph node metastases only seen on CXCR4 PET. G, H A 76-year-old patient with initial diagnosis of a gastric NEC and metastases in lymph nodes, liver, bones, and the lungs

CXCR4 PET-Derived Tumor Volume Remains a Prognostic Marker in Multivariable Cox Regression

Multivariable cox regression including age, Ki67-score, presence of liver metastases, and pretreatment (yes vs. no) for CXCR-derived TV and TLU confirmed TV as a significant predictive marker as well as for OS and for rPFS (OS: HR 1.012 95% CI 1.003–1.022, p = 0.0084; rPFS: HR 1.009 95% CI 0.9999–1.019, p = 0.0491), whereas TLU remained only predictive for OS (HR 1.009 95% CI 0.9999–1.019, p = 0.0194) but narrowly failed significance for rPFS (p = 0.0559). The other parameters were not significantly associated with OS or rPFS (p > 0.07, respectively).