FDG Uptake Predicts Recurrence and Survival in Patients with Unresectable Stage III Non-Small Cell Lung Cancer treated with 74 Gy (RBE) Proton Therapy and Chemotherapy

Reviewer: Abigail Berman Milby, MD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: October 4, 2011

Presenting Author: J. Chang
Institution: M.D. Anderson Cancer Center, HOUSTON, TX


  • Concurrent chemoradiation is the standard of care for locally advanced, unresectable non-small cell lung cancer (Curran et al., RTOG 9410, JNCI, 2011).
  • Chang et al. (Cancer, 2011) recently presented a phase 2 study of high-dose proton therapy with concurrent chemotherapy for unresectable stage III nonsmall cell lung cancer. They found that proton therapy was well-tolerated with no grade 4 or 5 adverse events and minimal grade 3 toxicity. The overall survival and progression-free survival rates were 86% and 63% at 1 year.
  • Standardized uptake value (SUV) is a ratio, calculated by regional radioactivity divided by total body radioactivity.
  • There have been several publications that show maximum SUV (SUVmax) from [18F] fluoro-2-deoxy-D-glucose (FDG) positron emission tomography/computed tomography (PET/CT) predict outcomes after radiation alone or chemoradiation, including a meta-analysis by Berghmans et al. (J Thor Oncol, 2009).
  • Proton therapy, due to its characteristic Bragg peak, may be advantageous in radiation for lung cancer by its ability to spare more normal tissue including lung and heart.
  • The goal of this study was to determine if SUVmax predicts time to local recurrence (LR), distant metastasis (DM), progression-free survival (PFS), and overall survival (OS) after concurrent proton therapy and chemotherapy for unresectable stage III non-small cell lung cancer (NSCLC).


  • Eighty-four (84) consecutive patients were enrolled and treated with 74 Gy(RBE) passive scattering proton therapy with carboplatin and paclitaxel chemotherapy.
  • All patients underwent 4D-CT-based treatment simulation and re-simulation during proton therapy for potential adaptive re-planning.
  • 18F-FDG PET/CT scans were obtained from all patients within 1-6 months before (SUV1) and after chemoradiotherapy (SUV2)
    • Delta (?) SUV was calculated by subtracting SUV2 from SUV1.
    • SUV cutoff values were identified by receiver operating characteristic analysis.
  • Univariate, utilizing Gray's test, and multivariate Cox regression analyses, utilizing Fine and Gray regression were used to determine the ability of SUV2, ?SUV, and other clinicopathologic factors to predict LR, DM, PFS, and OS.


  • Median follow-up time was 19.2 months (range 6.1-52.4 months).
  • The median survival time was 29.9 months, and 48 patients (57%) were alive at the time of analysis.
    • Local recurrence (LR) occurred in 14 patients (16.7%). Seven (7) patients (8%) had lymph node recurrence outside the planning target volume.
    • Distant metastases (DM) occurred in 39.3% of patients (crude rate).
  • SUV1 cutoff value of 14.2 (P=0.021) was found to predict DM-free survival (DMFS)
  • SUV2 cutoff value of 3.6 (P=0.033) was found to predict for LR-free survival (LRFS).
  • On univariate analysis, out of 12 clinicopathologic features evaluated (age, sex, performance status, smoking status, tumor histology, gross tumor volume, lung V20, mean lung dose, SUV1, SUV2, ?SUV and ?SUV/SUV1):
    • SUV2 (P<0.001) and delta(SUV) predicted LRFS
    • SUV1, SUV2, and performance status predicted DMFS, PFS, and OS (P=0.049~0.001).
  • On multivariate survival analysis:
    • SUV2 (P=0.006) was independently prognostic for LRFS
    • SUV1, SUV2, and performance status were independently prognostic for DMFS and PFS, and OS (P=0.035~0.010)

Author's Conclusions

  • High-dose proton therapy with chemotherapy produced a cumulative local control rate of 80.5% at 2 years and 29.9 months median survival in patients with unresectable stage III NSCLC.
  • SUV2 was an important independent prognostic factor for LRFS; both SUV1 and SUV2 predict DMFS, PFS and OS.
  • Delta (?) SUV predicts for time to local recurrence.
  • Understanding the SUV1, SUV2, and delta (?) SUV may help elucidate which tumors are more aggressive and may benefit from higher doses of radiotherapy or additional chemotherapy.

Clinical Implications

  • This study shows that SUV1, SUV2, and the difference between them can be predictive of different outcomes following chemoradiation with proton beam therapy for locally advanced lung cancer.
  • While the data is statistically significant, the clinical utility of this data is unknown. The published data regarding the predictive use of SUV(max) for outcomes in early stage lung cancer outcomes is varied, with some studies finding SUV(max) to be predictive and others not. Further studies will help confirm or refute the data presented here.
  • The authors specifically state that one PET/CT scanner, identical amount of isotope, and algorithm for determining SUV(max) was used. However, PET/CT equipment at different institutions is widely variable and the absolute value of SUV(max) is not consistent among scanners. Therefore, the absolute cutoffs given here for SUV1 of 14.2 and SUV2 of 3.6 cannot be broadly applied without further external validation.
  • The timing between SUV1, chemoradiation, and SUV2 is not fixed in this study (1-6 months before and after chemoradiation was permitted). The authors could explore if the timing of SUV1 and SUV2 affects the predictive quality of the SUV(max).
  • The local control reported here is over 80% at 2 years, which is significantly better than historical controls; it will be interesting to see this long-term data mature as it is very promising for including proton therapy in the definitive treatment of locally advanced lung cancer.