The Pill That Could · What Cure Means

What Cure Means, part 2

The first IRIS readout, published in the New England Journal of Medicine in March 2003, was decisive. At eighteen months of follow-up, the rate of major cytogenetic response — a substantial reduction in Philadelphia-positive cells — was 87 percent in the imatinib arm versus 35 percent in the interferon arm. The rate of complete cytogenetic response was 76 percent versus 14 percent. The rate of progression to accelerated phase or blast crisis was about 3 percent versus 9 percent. By every endpoint the trial was measuring, imatinib was so much better than the previous standard that further accrual to the comparison was, in a strict sense, untenable. The trial's data monitoring committee allowed patients on the interferon arm whose disease had not responded adequately to cross over to imatinib, and most of them did.

IRIS continued to follow patients for years afterward, with updates published periodically. By the five-year readout in 2006, the picture had stabilized: imatinib produced complete cytogenetic responses in roughly four out of five chronic-phase patients, the responses were durable in most patients who achieved them, and progression to advanced disease was rare. The rate of progression in patients who had achieved complete cytogenetic response by twelve months was, by year five, very low — single-digit percentages of all such patients had progressed.

The ten-year IRIS update, published in 2017, extended the picture in the direction the early data had suggested. Among the original imatinib-arm patients, overall survival at ten years was about 84 percent. Survival free of CML, specifically — patients still alive with their disease in chronic phase or in remission — was higher than that. Most of the deaths over the ten years had been from causes unrelated to CML: heart disease, other cancers, the things people in their sixties and seventies die of. CML itself, in the imatinib-treated cohort, had killed only a small fraction of the patients enrolled, the great majority of whom had progressed during periods when the drug couldn't be optimized fully — patients who had developed early resistance, intolerance, or who had not been able to maintain adequate dosing.

These numbers represented a transformation. CML in the pre-imatinib era had killed roughly two-thirds of patients within ten years of diagnosis. CML in the imatinib era was killing roughly one-sixth, and most of that one-sixth was made up of patients whose disease had presented in advanced phase, or who had developed early resistance, or who had been unable to take the drug for one reason or another. For the patients on imatinib doing well, ten-year survival was approaching the survival of an age-matched general population. By the fifteen and twenty-year readouts that have come since, that approach has continued. The patients on the drug, doing well, are dying at roughly the rate that people of the same age without CML are dying. They are not dying of CML.

This is what happens, in clinical statistics, when a fatal disease becomes a chronic one. The mortality of the disease falls off; the mortality of everything else stays where it was; the survival curve of patients with the disease starts to track the survival curve of the general population. By the long-term IRIS data, the imatinib-treated CML population had effectively merged, in survival terms, with the population that had never had CML.

* * *

What this did not establish, however, was that any individual patient was cured.

The IRIS data showed that patients on imatinib were not dying of their CML. It did not show that the CML was gone. The deepest molecular tests — PCR-based assays that could detect the BCR-ABL fusion gene at levels of one in tens of thousands of cells — mostly continued to find some small signal, in most patients, even after years of apparent remission. The disease was suppressed deeply enough to be clinically silent, but it was, at the level of molecular detection, generally still there.

It is worth taking a moment on what the molecular tests actually measure, because the rest of this chapter rests on them.

When chapter six introduced the response ladder — hematologic, cytogenetic, molecular — it deferred a careful explanation of the third rung. We can take it now. Molecular response in CML is measured by quantitative PCR, a technique that uses small DNA primers to specifically amplify and count the BCR-ABL fusion transcript in a blood or marrow sample. The transcript is the messenger RNA produced by the fusion gene, the working copy that the cell uses to make BCR-ABL protein. PCR can detect this transcript with extreme sensitivity — down to roughly one transcript per ten thousand or hundred thousand normal transcripts of a control gene, depending on the assay's design and the sample's quality. Each result is reported on a logarithmic scale: a major molecular response (MMR) is a thousandfold reduction from a standardized baseline, conventionally written as MR3.0; a deep molecular response is ten or a hundred thousandfold (MR4.0 or MR4.5); a complete molecular response is undetectable BCR-ABL on a sufficiently sensitive assay (MR5.0 or beyond).

On these scales, most patients on imatinib in deep remission sit somewhere between MR3 and MR4 — still detectable, but at very low levels. A subset of patients, particularly those on more potent second-generation drugs or on imatinib for many years, achieve MR4.5 or deeper. Some achieve undetectable disease on the most sensitive assays available. The depth of molecular response has turned out to predict, more accurately than any other clinical variable, what happens if the patient stops the drug.

All 13 chapters of The Pill That Could · Library · All courses