When finance, not science, is the bottleneck
Oncology has entered an era in which it is “costlier to discover, harder to develop, and less rewarding to succeed.” Dr. Khozin’s work starts from that premise: many promising therapies fail not because the biology is wrong but because the capital structures around them do not fit the risk, cost and timelines of drug development. His work identifies four structural failures and engineers financing to correct them.
Risk–return misalignment
Individual cancer programs can fail more than 95% of the time, a risk profile conventional investors will not hold.
The valley of death
Promising preclinical science stalls before first-in-human trials, too early for most investors and too costly for academia.
Short-term capital, long-term science
Development takes 10 to 20 years, while most venture funds have ten-year lives and seek returns within five to seven, forcing exits midway.
Misallocated investment
Capital chases fashionable areas in boom-and-bust cycles, leaving high-need cancers underfunded.
The economics of a single program
A drug program is worth funding when the profit it is expected to earn, valued in today’s dollars, exceeds what it costs to develop. Net present value (NPV) puts both in a single sum: every year’s net cash flow, discounted back to the present.
- CFtNet cash flow in year t
The program’s profit in year t: revenue minus every cost. During development it is negative, because the program spends money and earns nothing; after approval it turns positive as sales arrive.
- rDiscount rate
The annual return investors require for the risk they take, also called the cost of capital, written as a decimal (0.10 for 10%). At r = 0.10, $1 earned ten years from now is worth about 39¢ today.
- tYears from today
How far in the future a cash flow arrives. Each year’s cash flow is divided by (1 + r) once for every year it lies ahead, so distant profit counts for less than near-term profit.
- TLast year of commercial life
The final year the drug earns, usually when its patents expire and generics enter. It equals the years spent in development plus the years on the market.
- NPV > 0The threshold
A positive NPV means the program returns more than its cost of capital and is worth funding. Below zero, rational investors pass, however promising the science.
Years 1 to approval, when CFt is negative. Shortened by streamlined trials and real-world data.
Spending on preclinical work and clinical trials.
CFt after approval. Raised by pricing, reimbursement and market access.
From approval to patent expiry. T is development plus market years.
Lowered when risk can be pooled and shared, as in the frameworks below.
- Development, years 1–8
- −$320M
- Market, years 9–20
- $1.1B
Novel financing frameworks
Dr. Khozin’s work focuses on the following financing structures and how they can be applied to correct those failures. Select one for how it works and how it helps.
-
How it works
A single vehicle holds a large, diversified portfolio of development programs across stages, targets and mechanisms. It is financed with a layered capital structure: equity at the bottom and tranches of research-backed obligations, bonds repaid from licensing, milestone and royalty income, above it. Because programs that are weakly correlated rarely fail together, the portfolio’s volatility falls far below that of any single asset.
How it helpsTurns assets too risky for any one investor into an instrument that pension funds and insurers can hold, lowering the cost of capital, r in the formula above.
-
How it works
Funds with 15- to 20-year terms, modeled on infrastructure finance, raised from patient capital such as endowments, sovereign funds and insurers. Income from later-stage assets or royalties can support the fund over its life, so early programs are not abandoned when a conventional 10-year venture fund must return capital.
How it helpsRemoves forced exits midway through development and aligns the duration of capital with the 10 to 20 years a cancer therapy takes to reach patients.
-
How it works
Foundations and patient advocacy organizations fund development programs directly, structured so that success returns royalties or milestone payments to the funder. Those returns are reinvested in the next programs, as the Cystic Fibrosis Foundation did in backing the therapies that transformed its disease.
How it helpsDirects capital to high unmet need that markets underweight, such as pancreatic and esophageal cancer, and creates an evergreen pool of funding.
-
How it works
Government co-investment, first-loss capital, guarantees or advance commitments that absorb part of the downside of early-stage research, alongside private investors who take the remainder.
How it helpsReduces the effective risk and duration private capital must bear, drawing investment into the gap between promising preclinical science and first-in-human trials.
-
How it works
A single multi-compound entity in-licenses promising assets from many universities and medical centers and builds the development infrastructure around them, so that scientists can remain in academia while their discoveries advance.
How it helpsConverts stranded academic intellectual property into an investable portfolio, sharing risk across compounds and lowering development cost through shared expertise.
-
How it works
Methods from quantitative finance and machine-learning models trained on clinical-trial data that estimate each program’s probability of technical and regulatory success, and how outcomes correlate across a portfolio. Dr. Khozin’s research with Andrew Lo combined machine learning with tumor-growth models to predict outcomes in advanced lung cancer.
How it helpsGives investors a defensible estimate of the odds of success, the input every financing structure depends on, so capital can be priced to the real risk rather than to uncertainty.
CEO Roundtable on Cancer, New York
One example of this work: a roundtable Dr. Khozin designed and led as CEO of the CEO Roundtable on Cancer, bringing leaders from science, finance and policy together around these failures and the frameworks to address them, with a keynote by MIT’s Andrew Lo.
