A three-marker blood test flagged 87% of stage 1 and 2 pancreatic cancers in 1,800 patients

The false-positive rate scales with who you test. In this trial the high-risk figure was more than five times the low-risk one ^[5][26]. High-risk people are the ones a screening programme enrols first. Discover Magazine reported that the accuracy gained from combining three biomarkers should especially benefit people at high risk ^[20][33], and Scientific American reported that doctors would use the score particularly in people with inherited risk or a history of pancreatitis ^[24]. There is no routine screening test for pancreatic cancer the way there is for breast or colon cancer ^[14]. A flagged patient goes on to further evaluation ^[34].

How tolerable that burden is depends on how many people in the screened group actually have cancer. Neither account reports how many cancers were found among the 1,800 enrolled ^[2][3]. The positive predictive value of a flag cannot be computed from what has been published. The 87 percent covers stage 1 and stage 2 pooled ^[4].

The test also misses cancers. About 13 of every 100 stage 1 and 2 cancers in the trial went unflagged ^[27]. For high-grade dysplasia, the precancerous state found in people with high-risk pancreatic cysts, detection ran above 64 percent, meaning under 36 percent were missed ^[7][8][31]. Discover reported that identifying dysplasia could help physicians monitor cysts and decide which ones might require intervention ^[21].

Ajay Goel, the study's senior author and chair of molecular diagnostics and experimental therapeutics at City of Hope, framed the result around timing ^[10]. "For patients, these findings represent progress toward finding pancreatic cancer before symptoms appear and while more treatment options remain available," he said in a statement ^[11]. "A stage shift is not just a statistic," Goel said ^[12].

Late-stage treatment shows what a stage shift would be worth. The Food and Drug Administration has approved daraxonrasib for people with late-stage pancreatic cancer who have already been through chemotherapy ^[17]. It puts survival at 13.2 months against 6.7 months for chemotherapy alone, a gain of 6.5 months ^[18][28]. Five-year survival after diagnosis is 13 percent in Scientific American's account and 14 percent in Discover's, which cites the National Pancreatic Foundation ^[15][16].

The enrolment was an early-stage clinical trial, and Scientific American reported that more advanced trials are needed before the test could be used in the clinic ^[3][19]. The two publishers also spell it differently: Discover writes PANXEON, Scientific American writes PAXEON ^[29][30].