A 72-year-old man presented for evaluation of progressive dyspnea and cough. CT of his chest revealed bullous emphysema (bottom arrow), a tumor involving the middle lobe of the right lung (top arrow), and a pack of cigarettes in his shirt-pocket (asterisk). Biopsy of the lesion confirmed the presence of non-small-cell lung cancer. Source: N Engl J Med 2006; 354:397.

Although PET and CT both offered concordance with histologic measurements of non-small-cell lung cancer (NSCLC), CT is often accurate for staging primary T1 and T2 NSCLC, according to a study published in this month's American Journal of Roentgenology.

The advent of IMRT (intensity-modulated radiation therapy) for NSCLC has strengthened the imperative for accurate staging of the primary NSCLC tumor. However, “radiotherapy planning with CT is inaccurate, definition of tumor volume being the main source of error,” explained Davina Pawaroo, MD, of the department of radiology at Norfolk and Norwich University Hospital in England. PET/CT provides improved staging of NSCLC and of the primary tumor, offered Pawaroo.

Pawaroo and colleagues designed a retrospective study to compare the maximum measurement of T1 and T2 primary NSCLC on CT and PET images to determine which modality best correlated with the reference standard--histologic measurements.

The study focused on the analysis of PET/CT image sets acquired between 2005 and 2007 and histologic measurements for 59 NSCLC patients.

A single radiologist measured the dimensions of the tumor in the maximal plane and also measured the tumor on unenhanced CT images with soft tissue and lung windows. A second radiologist reviewed the measurements.

The analysis revealed the highest correlation between CT with soft tissue windows and CT with lung windows with all variables showing significant positive correlation with the maximum measurement of the histologic specimen. Although soft-tissue windows showed the most concordance with histologic measurements, the maximum measure on PET datasets had a smaller standard deviation, indicating less variation in measurements.

Several tumors did not follow the general pattern of others. Three tumors containing an adenocarcinoma with a bronchoalveolar growth pattern had maximum PET measurements less than the histologic measurements because part of the tumor had an SUV less than 2.5. The authors suggested that employing PET for measurements of tumors of this histologic type could lead to underestimations. In addition, “two tumors with an SUV less than 2.5 were unmeasurable with PET,” wrote Pawaroo.

However, the authors did acknowledge that PET or CT alone provides sufficient accuracy for measuring most primary lung tumors.

The researchers outlined several challenges with PET imaging: wide variability in maximum SUV of primary NSCLC tumors, the lack of a standard for placing contours around the tumor on PET scans and poor delineation between gross tumor volume and single maximum SUV.

Five patients with consolidation or collapse surrounding the tumor had CT measurements greater than in the histologic specimen, “indicating that CT is less accurate in these specimens.” In such cases or those with possible invasion of the mediastinum, the authors indicated PET may be useful for delineating the primary tumor volume.

Pawaroo and colleagues noted that the use of pathologic specimens was a shortcoming of the study because the formaldehyde used to preserve the specimens causes shrinkage. Also, the researchers did not have data about the original plane in which the specimen was measured and suggested a prospective study in which specimens are measured in a stated plane after resection.

A second area for additional research is an investigation of whether measurements would differ depending on the presence of the tumor in the upper or lower lobe of the lung, offered the researchers.