This article was originally published on Fortune.com, by Sy Mukherjee May 26, 2016.

But there’s good reason to hope that may change very soon.

This post was corrected/updated on Monday, May 30.

A massive new study of drug candidates sheds light on which experimental therapies are most likely to eventually reach the market.

The report was compiled by biotech’s largest trade association, the Biotechnology Innovation Organization (BIO), with the help of business intelligence firms BioMedTracker and Amplion, and it found that drugs for blood disorders have the highest probability of ultimately winning FDA approval. Meanwhile, cancer drugs have the lowest chance of receiving the go-ahead.

BIO’s David Thomas, the report’s author, noted that the study is the largest-ever of its kind in an interview with Fortune. The study examined a decade’s worth of data (2006-2015) encompassing just under 10,000 “phase transitions,” a term for an experimental therapy’s progression through three phases of human clinical trials, regulatory filing, and ultimate approval.

On average, a drug candidate had a 9.6% chance of going from phase I trials to winning the FDA’s blessing, according to the study. Transitions from phase I to phase II were the most successful—not too surprising given that phase I trials only assess safety and not effectiveness. Progressing from small phase II efficacy studies to far larger phase III ones proved the most challenging.

On average, a drug candidate had a 9.6% chance of going from phase I trials to winning the FDA’s blessing, according to the study. Transitions from phase I to phase II were the most successful—not too surprising given that phase I trials only assess safety and not effectiveness. Progressing from small phase II efficacy studies to far larger phase III ones proved the most challenging.

 

probability of drug approval
BIO

So, which therapy types fared best? Hematology, or blood disorder medications, were the clear winners, with a more than 26% chance of final approval. Infectious diseases and ophthalmology (eye diseases) meds rounded out the top three with 19.1% and 17.1% success rates, respectively.

Cancer drugs were by far the least likely to reach market, with a 5.1% chance of approval (psychiatric and cardiovascular medications also ranked in the bottom three at 6.2% and 6.6%, respectively). Oncology was also the clear outlier across the 14 examined disease types when it came to the probability of a successful phase III. Just 40% of these therapies produced results strong enough to warrant filing an approval application at that stage, while six other disease categories had more than 70% chance of succeeding.

Click here to subscribe to our new Brainstorm Health Daily Newsletter.

On the flip side, 79% of cancer therapies that made it all the way to the application stage were approved on the first try—the highest first-attempt approval rate of any therapeutic space.

The study also found that highly-personalized drug trials, which examine a relatively homogenous patient pool, have a significantly better chance of success compared to those focused on more prevalent diseases that affect a wide range of people. For instance, rare disease drugs had a one-in-four shot of making it from phase I to approval, nearly triple the rate of success for common conditions like cardiovascular diseases.

rare disease vs other drugs
BIO

That shouldn’t come as much of a surprise. After all, therapies interact with different people in different ways, so the more similar a target population, the easier it is to create a therapy tailored to treat its disease (the FDA also gives priority to medications for rare disorders).

But this also speaks to a broader trend in biopharma that could help shift the lackluster approval rate of cancer drugs. Drugmakers are increasingly making use of “biomarkers,” telltale biological signs that distinguish certain patients from others. That might mean the elevated presence of a certain protein or the existence of a specific gene in a patient which makes that person more or less likely to have a disease type or respond to a treatment.

According to the report, biopharma companies that used biomarkers as a criteria for either including or excluding trial participants were three times more likely to win approval than those that did not.

biomarkers approval likelihood
BIO

It’s also further evidence of the industry’s shift away from developing drugs for broad, heterogeneous populations to more personalized therapies. “I think that trend started, for me, in 2011,” said BIO’s Thomas. “That’s when [French drugmaker] Sanofi bought [the biotech firm] Genzyme. And what’s influencing the rate of success for both [rare diseases and biomarkers in clinical trials] is really the science.”

“From the biomarker standpoint, the science is uncovering new genetic markers that are tied to the disease,” Thomas added. “A lot of that comes from the gene sequencing that we can now do for much cheaper and has been done on hundreds of thousands of people’s genomes now.”

Pharma giants like  BMY 1.57%  and Merck  MRK 0.87%  have already experienced major success with their investments in personalized cancer treatments like Opdivo and Keytruda (the latter drug has been credited with clearing President Jimmy Carter’s cancer). And the BIO report suggests that the growing use of biomarkers and interest in immunotherapies could significantly boost cancer drug approvals down the line.