409 ‒ Inside modern drug development: the science, economics, and regulatory hurdles

Lloyd describes how he and his team built a list of about 7,000 clinical indications, grouped them into what Novartis was already doing, rare developmental disorders, and seven or eight unmet buckets such as healthy aging, ENT, renal disease, and fibrotic diseases. This unit eventually grew dozens of projects and became a monster. He personally prefers large indications because he has limited time and wants to help as many people as possible, even though smaller orphan indications can have faster regulatory paths.

Lloyd says every successful program has almost been killed several times. He learned that toxicity must be monitorable and reversible; serious irreversible organ toxicity is usually the end. He warns that even late-stage success can be a commercial failure if nobody needs or pays for the drug, so the risk filter is not just FDA approval but real-world value creation.

Lloyd explains he likes to dose as high as possible to understand human tolerability while never exceeding exposures tested in animals. In bimagrumab's case, they went up as high as they could and the answer was nothing, meaning no serious adverse event attributable to overdose. The animals had been dosed weekly and exposures were intentionally higher than expected human exposures to create a safety margin. In oncology, toxic and therapeutic exposures can overlap and patients accept the risk because medical need is great.

Lloyd says the first thing companies do for drugs with abuse potential is work with WADA. For bimagrumab, as soon as the mouse antibody worked, they started that process, and WADA has had an assay for more than ten years. Asked whether WADA has ever caught anyone using it, he does not know.

Lloyd's cancer-prevention company is based on sorafenib-associated skin cancers. He explains that a Phase 2 can actually measure prevention in high-risk people because the baseline event rate is so high. For systemic cancer prevention, incidence is too rare to test in healthy people, but adjuvant prevention in high-risk patients with no evidence of disease is a possible path. The Broad Institute panel showed melanoma was the most sensitive tumor type in cell lines, so he is confident the mechanism may also treat melanoma.

Lloyd separates drug discovery into incremental improvements (less frequent dosing, oral over injectable, or a better version of a known drug) and quantum-step new therapies for indications that may not even have ICD-10 codes. The industry is biased toward incremental because risk and uncertainty are low; the heavier lift requires creating awareness, a regulatory path, and eventually payer acceptance. He chose large indications for personal reasons: limited time, wants to help as many people as possible. For sarcopenia, he argues frail elderly people were a much bigger medical need than classic orphan diseases, even though orphan indications offer faster regulatory pathways.

Lloyd designed a two-part study: first, validate the accelerometer against videotaped falls of ice skaters in Boston, then deploy it in nursing-home residents where a person found on the floor was the positive control. Funders cut the ice-skater validation, so he worked with geriatrician Lou Lipsitz and ran the nursing-home study. The device detected only 17% of real falls and had a very high false-actuation rate, so it couldn't measure falls or be used as an endpoint. They also explored Dina Katabi's RF/Wi-Fi sensing but could not use it, and ended the falls-prevention drug program.

In rodents, bimagrumab-like antibodies were more impressive than myostatin knockouts: huge muscle, stronger, faster. In humans, the mass gain was 4–8% max, with trends to better responses in males and younger adults, but strength and performance did not improve in a major way. A meta-analysis of Novartis sarcopenia studies showed a 9-meter increase in six-minute walk distance, which Lloyd doubts will prevent falls. He notes this same pattern—larger muscle without training-induced strength—occurred with IGF-1 agonists, androgen agonists, and SARMs. The biology appears substrate-dependent: more protein intake produced more muscle in a nutrition substudy.

Versanis repositioned to study bimagrumab on top of GLP-1 therapy, ran a full factorial 9-arm study to see dose responses and combination effects. Semaglutide was open-label because the commercial auto-injector couldn't be placebo-controlled; bimagrumab was IV. Primary endpoint was body weight, chosen for regulatory and investor reasons, though Lloyd wanted waist circumference. DEXA was done in every patient. Results: high-dose combination lost 22–23% body weight and 45.7% of starting body fat by 72 weeks. Grip strength improved only slightly. Lloyd calls this the first medical therapy with fat loss equivalent to or better than bariatric surgery.

Lloyd says the LDL rise is a direct effect of the drug in the liver, not an off-target chemical effect. It was about 20%, and it matters because LDL predicts adverse cardiovascular endpoints. But he argues it is monitorable and treatable, which is why he would have continued into Phase 3 if Versanis had remained independent. Chris Lu, who later founded a company making ActRIIA/B antibodies, likely knows the exact biology but hasn't published it yet.

mTORC1 integrates nutritional input and switches between growth and autophagy. In young rodents, fasting downregulates mTORC1; in old rodents it did not. That makes Lloyd question whether intermittent fasting will cause the same metabolic switch in older people. He notes that rapalogs are mTORC1-selective initially but downregulate mTORC2 with sustained exposure. Restore Bio tried combining a catalytic and allosteric inhibitor; other companies are trying to make a real mTORC1-selective inhibitor. He emphasizes the biology is hard to measure in healthy humans because you can't easily get liver tissue, unlike cancer trials where the pathway is highly upregulated.

If inhibiting the pathway causes cancer, the pathway is normally protective; Lloyd's innovation is to turn it on gently and controllably. Skin cancer is nearly as common as all other cancers combined, so that is the first target. A Phase 2 could enroll 100–120 adults age 50 and older with at least five prior skin cancers, who have a 50% chance of another within a year. In a Broad Institute screen of 1,000 cancer cell lines, melanoma was by far the most sensitive tumor, which Lloyd didn't expect: mutational burden correlates but doesn't fully explain it. He still emphasizes avoiding carcinogens, treating obesity, and using sunscreen—those are not alternatives.