PET scans often inaccurate; may deny curative surgery for lung cancer
The use of positron emission tomography -- better known as PET scans -- has grown dramatically over the past 15 years, thanks to their seemingly magical ability to identify foci of undetected metastatic cancer.
But PET scans' perceived high accuracy in diagnosing metastatic non-small cell lung cancer (NSCLC) -- a published 94% sensitivity and 83% specificity in a 2001 JAMA meta-analysis -- may be vastly overstated, and mask wide geographic variability, according to more recent data. False-positive PET scans may be rampant in Histoplasma-endemic areas, and could result in misdiagnosis of metastatic lung cancer -- with denial of potentially curative resections for stage 1 NSCLC, emerging data suggest.
A 2011 study in Annals of Thoracic Surgery showed that 60% of PET scans suggestive of lung cancer in a Histo-endemic area (Nashville, TN) were false positives. Another 2011 review in the Journal of Thoracic Oncology showed that PET scans can wrongly diagnose locally metastatic disease (N3 lymph nodes) as often as 35% of the time.
Most recently, a post hoc analysis of 682 patients with suspected stage 1 lung cancer who all underwent surgical resection (from a data set presented at the 2012 American Society of Clinical Oncology/ASCO meeting) from dozens of U.S. cities added to the confusion:
Although 82% of the 566 confirmed cancers were PET-avid, 69% of the benign lesions were also PET-avid.
PET's positive predictive value was 85%, but its negative predictive value (likelihood that a negative PET was a true negative) was only 26%.
PET's specificity (avoidance of false positives) was low, and varied widely between cities: it was 46% in Philadelphia (a ~50% false positive rate) but only 15% in Birmingham, AL -- where 85% of the lesions resected were benign (this data suggests).
PET's overall accuracy was 73%, its sensitivity 82%, and its specificity only 31%.
In this series, granulomatous disease was the most common histology found on microscopy of benign lesions.
The National Comprehensive Cancer Network's guidelines recommend the routine use of PET scanning for the diagnosis of non-small cell lung cancer, but this was based on the previously accepted high-performance characteristics. Although no expert body has stepped forward to officially wave the caution flag around PET scanning, it seems evident that false positive PET scans could harm patients through either unnecessary resections or denial of curative resection from falsely positive lymph nodes and/or distant granulomatous disease. Harm from false negatives is also possible, but less likely, given the high sensitivity of PET scanning (although probably not as high as previously believed).
Wondering what your state's rate of Histoplasma infection might be, and how it might impact the performance characteristics of PET scanning where you practice? Here's a map from the Centers for Disease Control and Prevention (CDC) quantifying Histoplasma endemicity in the U.S.:
Geographic distribution of histoplasmosis in persons >65 years of age, United States, 1999–2008. Values are no. cases/100,000 person-years. Source: CDC.
Originally reported in Chest Physician, August 2012 issue.