267 ‒ The latest in cancer therapeutics, diagnostics, and early detection | Keith Flaherty, M.D.

KEYPOINTS

• 01

  Cancer's share of deaths peaks in midlife

  By age band, cancer causes about 6% of deaths at 25–34, 13% at 35–44, 23% at 45–54, 30% at 55–64 and 31% at 65–74. The share then falls only because heart disease and neurodegeneration take over. Cancer ranks among the top three causes of death in every decade from the mid-20s and is number one in parts of midlife. Five cancers (lung, breast, prostate, colorectal, pancreatic) account for just over half of US cancer deaths.

• 02

  Hormone-driven cancers appear when the hormones are low

  Peter points out that men have their highest testosterone and DHT in their 20s to 40s, yet prostate cancer appears when levels have fallen. Breast cancer likewise doesn't peak at peak estrogen. Low-androgen prostate cancers tend to be worse, as a New England Journal paper showed about 15 years ago. Keith Flaherty's takeaway is that the biology runs deeper than 'hormone in, cancer out'.

• 03

  Obesity is the second-biggest environmental cause of cancer after smoking

  About 25–27 cancers have a tight link to obesity. Peter suspects the driver is not fat itself but insulin, probably IGF-1, and the inflammation that comes with it. These effects take years to kick in, so rising obesity rates today predict more cancer later.

• 04

  Survival with metastatic solid tumours barely moved for 30 years, and has only recently started to

  Ten-year survival with distant spread was essentially 0% in 1970 and about the same in 2000. The only exceptions were the 10–15% of kidney cancer and melanoma patients who responded to high-dose interleukin-2. Flaherty thinks 15% is now a conservative estimate for durable survival, moving toward 20%. He notes that 52 FDA approvals since then cover only 19 unique mechanisms, and roughly half the gain comes from the PD-1 drug Keytruda.

• 05

  PD-1 does the heavy lifting in immunotherapy

  Tumours display PD-L1, 'the foot that presses on the brake' of T cells. Blocking PD-1 gives roughly 10% of all cancer patients long-term, roughly 10-year-type benefit, by Flaherty's math. Adding CTLA-4 blockade helps about a fifth of tumour types but contributes only around 1% in addressable patients on top of that.

• 06

  Most epithelial tumours are visible to the immune system

  Steve Rosenberg's finding that roughly 80% of epithelial tumours carry novel neoantigens means the immune system could, in principle, take most cancer deaths off the table. Spontaneous regressions are rare but real: Rosenberg's career began after seeing a patient who had been sent home to die with metastatic gastric cancer return 10 years later, needing his gallbladder out and cancer-free.

• 07

  Cell therapies work but are hard to scale

  CD19 CAR-T is the poster child: every B cell carries CD19, so lymphomas arising from B cells can be targeted. Most solid tumours lack such a clean target, so researchers are engineering cells that require two markers. TIL therapy requires surgically removing tumour, for example a wedge of lung, then growing its lymphocytes to about 10⁹ cells and reinfusing them with interleukin-2. That is costly and bespoke, and it often fails against the tumour's defences.

• 08

  Tumours build a 'force field' that has to come down first

  Flaherty uses a Star Wars analogy: you have to take out the shield generator before attacking the Death Star. Cancers recruit suppressive immune cells and fibroblasts (notably in pancreatic cancer) and alter their own metabolism. Lasting cures will need combinations matched to each patient, not one cocktail. Peter asks what it would take, over five years, to double long-term survival with these therapies from about 20% to 40%, a number he says he is making up for illustration.

• 09

  The same drugs cure microscopic disease but not visible metastases

  Stage 3 colon cancer treated with surgery and chemotherapy gives about 60–70% five-year survival. Add liver metastases and give the same drugs, and five-year survival falls to somewhere between none and a few percent, with none alive at 10 years. Flaherty's explanation is clonal: drugs kill the dominant 99% of cells, but the remaining 1% contains about 10 subclones, one of which already carries resistance. Larger tumours hold more of them. Hence his view that early detection is the problem that 'wins the game'.

• 10

  Liquid biopsies are reasonable now and will get much better

  Blood tests such as Grail's (ordered by a physician) detect tumour DNA and suggest the tissue of origin. Flaherty considers testing reasonable for early adopters who can tolerate uncertain results. Peter uses Grail selectively. Methods in development are 10–100× more sensitive from the same 10–20 mL blood draw, and repeated testing over 6–18 months may be needed to locate a cancer after a positive result.