For patients with non-small cell lung carcinoma, volumetric data derived from quantitative PET imaging could provide more precision staging and inform physicians about patient survival, according to a study published in the March issue of the American Journal of Roentgenology.

Jessica Davison, MD, from the department of radiology at Boston University School of Medicine and colleagues, examined the numbers, using PET-based volumetric parameters as a means of assessing patient survival.

An estimated 20 percent of lung cancers are defined as non-small cell lung carcinomas (NSCLC). Conventional staging and therapy determination for NSCLC is based on evaluations of tumor size, presence of lymph node involvement and other metastases. CT-based methods of evaluating gross tumor volume are common, but they lack the physiologic information provided by PET. For this reason, F-18 FDG PET/CT is considered the imaging technique of choice for NSCLC.

“There is increasing interest in the utilization of functional rather than anatomic parameters in tumor volume delineation because incorporation of metabolic data may significantly improve determination of the biologically relevant lesion volume and even take into account the heterogeneity within the tumor itself,” wrote Davison et al.

Davison and colleagues focused on quantitative PET data—specifically threshold and gradient-method PET segmentation of total glycolytic activity, metabolic tumor volume and maximum standardized uptake value (SUVmax). Researchers analyzed the data using overall patient survival as the main outcome measure. In order to assess tumor volume, researchers applied gradient PET segmentation algorithms using quantitative software, inputting operator-defined starting points localized within the lesion to find count-level variations along the tumor’s edge.

For this retrospective study, a total of 39 patients with a mean age 67.6 years diagnosed with NSCLC after PET/CT imaging between March 2007 and May 2009 with a 12-month or longer followup were selected via electronic archive. Researchers obtained PET segmentation of total glycolytic activity using the gradient method, metabolic tumor volume, and SUVmax and performed Kaplan-Meier curves for survival analysis as well as receiver-operating characteristic curve analysis and multivariate logistic regression. A comparative analysis was conducted using the Mantel-Cox log-rank test.

Results revealed that at the end of one-year followup, 19 patients had survived and 20 had died. NSCLC cases were categorized in cancer stages I-IV with 26 percent of the study group at stage I, 15 percent at stage II, 41 percent at stage III and 18 percent of subjects at stage IV NSCLC.

Those who died had a substantial uptick in the mean gradient-based total glycolytic activity and metabolic tumor volume (93.3 mL and 597.5 g) versus survivors’ data (19.3 mL and 193.9 g) for p values of 0.003 and 0.031, respectively. The difference between mean SUVmax for the two groups was unremarkable (p=0.85).

“The significant association between metabolic tumor volume and short-term prognosis (i.e. mortality within 12 months) probably reflects the increased mortality expected in patients with a large burden of growing tissue (metabolically active fraction),” wrote the authors. “For these patients, as long as the tumor grows beyond a certain rate, it is the volume of tumor at presentation that determines mortality within one year after diagnosis. The association between total glycolytic activity and 12-month survival may be related to disease burden as well as tumor aggressiveness.”

Investigators concluded that resulting values of gradient method of segmentation for total glycolytic and metabolic tumor volume were well matched to the threshold-based algorithms. Gradient-based algorithms for metabolic tumor volume were especially associated with one-year survival, indicating that these volumetric PET parameters could provide meaningful information about patient prognosis. Further studies are required in larger patient populations to corroborate these findings, the researchers added.