The PET Response Criteria in Solid Tumors (PERCIST), published in May 2009 in the Journal of Nuclear Medicine, aim to propel PET from the murky realm of qualitative imaging to a quantitative breakdown of response to cancer treatment. However, transitioning PERCIST from proposal to practice remains a work-in-progress. Health Imaging explores the rationale for PERCIST, benefits of the model and barriers to its dissemination.

PET is a quantitative modality, yet most reports are qualitative and descriptive, rather than quantitative, explains Richard L. Wahl, MD, director of nuclear medicine/PET at Johns Hopkins University School of Medicine in Baltimore. “We developed PERCIST because there is a lot of variability across practices in terms of PET interpretation.” PERCIST, along with related initiatives such as RSNA’s Quantitative Imaging Biomarker Alliance, aims to standardize PET protocols. The goal, says Wahl, is to transform PET from an imaging device to an imaging and quantitation device.

It turns out that variability in PET imaging may translate into a high downstream price tag. Current targeted cancer treatments, such as Gleevec, can cost $6,000 to $7,500 per month, explains Johannes Czernin, MD, from University of California, Los Angeles. Early knowledge of whether or not a drug is working for a given patient allows physicians to switch or discontinue an ineffective therapy. PERCIST may offer these data as early as after two cycles of chemotherapy, far more rapidly than the conventional months after treatment. “It’s a cost-savings and quality of life issue,” says Czernin, who emphasizes that appropriate use of PET can save money for the healthcare system and represents 1 percent of cancer costs.

Auto-PERCIST identifies disease and locates SUL-PEAK at 28 days post-therapy in PET study of breast cancer. Source: JP Leal and RL Wahl, Johns Hopkins

Precision drives PERCIST

PERCIST specifies explicit PET acquisition parameters, particularly the time from injection to imaging. PET practices typically image patients anywhere from 40 to 90 minutes after the FDG injection. That’s a problem because SUV uptake in the reference organ, typically the liver, changes over time, which affects tumor interpretation. It may not be possible to determine if a change in SUV is a random error or a true change in metabolic response when comparing a PET study acquired 40 minutes post-injection with one acquired at 90 minutes. “PERCIST in the clinic requires closer attention to the technical details than in routine practice, where qualitative interpretations are provided,” explains Wahl.

He and his colleagues envisioned that clinical trials first would adopt PERCIST to facilitate trials of drug development, and then as the approach was validated in literature, PERCIST could be applied to individual patients. However, shifting gears from existing anatomic tumor response metrics such as RECIST and the WHO criteria faces a few hurdles.

At Johns Hopkins, the nuclear medicine and radiology departments have tried to establish a foundation for PERCIST and improve adherence to uptake time recommendations, which has required tighter management by imaging technologists and fellows. The departments analyzed workflow to pinpoint and remove bottlenecks, such as patient scheduling, transport and injections, preventing tight adherence to the standardized PERCIST approach.

Stepping stones

One barrier to more widespread dissemination of PERCIST is the dearth of software that can measure PERCIST metrics. “Most providers have PET systems that are several years old, and the software cannot measure PERCIST inputs,” says Dominique Delbeke, MD, director of nuclear medicine and PET at Vanderbilt University in Nashville, Tenn. and past president of the Society of Nuclear Medicine and Molecular Imaging. The lack of software translates into a time-consuming measurement process, as PERCIST requires measurements of SUV in the liver, SUV peak and glycolysis.

One workaround, says Wahl, is to use tumor SUV max; however, the metric is somewhat more variable than SUV peak. On the plus side, software development seems to be catching up as several vendors have implemented PERCIST modules. Researchers at Johns Hopkins have developed Auto-PERCIST software to use in clinical trials for more expedited and accurate interpretation.

Wolfgang Weber, MD, chief of molecular imaging and therapy at Memorial Sloan-Kettering Cancer Center in New York City, who uses PERCIST in clinical trials, says software can play a key role in adoption. “Any PACS software that would store previous measurements and show images side-by-side and automatically copy measurements into the report could help to significantly shorten the time needed for these measurements.”

Another challenge is inter-vendor/system disparity as SUV on one system does not equate to others. Vendors are attempting to address the issue with software to increase accuracy.

The issue of time shouldn’t be overlooked, adds Wahl. “Time is money. We have to ask if any additional work is worth it. The jury is still out for PERCIST in routine practice.” Johns Hopkins applies PERCIST in a limited number of cases, especially those with multiple tumors and research has suggested that improved clinical outcomes for PERCIST responders, or a decline of 30 percent in SUV peak in the hottest lesion, correlates with good outcomes in a number of cancers, including lymphoma, sarcoma and esophageal and lung cancer.  

The decisions of early adopters, specifically clinical trial sponsors such as the National Cancer Institute, also will impact the promulgation of PERCIST, says Delbeke. These organizations may adhere to RECIST, or there may be a move to add a measurement of FDG uptake to RECIST.

“Will they want to adopt a completely new measure that has yet to be validated or remain with the older measurement?” One point in favor of familiarity is that any transition from one criterion to another negates the possibility of comparison. That is, response measured via RECIST cannot be compared with data provided via PERCIST. However, only by testing new methods like PERCIST can its validity be further determined, Wahl counters.

The final hurdle presents in the form of knowledge. PET providers need to be aware of the guidelines. “They need to read them and apply them,” says Weber. In addition, the referring physician community may be in the dark about the framework, and also requires education. At Johns Hopkins, Wahl and colleagues are laying the groundwork for an informed referral community by routinely reporting uptake time and uptake in the reference organ in structured, quantitative PET reports.

Ultimately, the PERCIST criteria provide more robust assessment of patient response than other measures, with the potential to allow a more precise definition of response, sums Delbeke.