How can the clinical course of advanced systemic mastocytosis on KIT inhibitors be predicted?

Two drugs, two different mutation panels

The paper's central claim is that the clinical course of advanced systemic mastocytosis (AdvSM) can be stratified before treatment starts, but not with one shared gene list. In the avapritinib cohort, the number of mutated genes among SRSF2, RUNX1, and SETBP1 sorted patients into three risk tiers. In the midostaurin cohort, the relevant panel was SRSF2, ASXL1, and EZH2. Each panel was weighted by hazard ratios estimated from that cohort, so the tiers are drug-specific rather than interchangeable.

The cohorts were not assembled the same way. The 176 avapritinib patients came from the phase 1 EXPLORER and phase 2 PATHFINDER trials, treated between 2016 and 2021, with 46% having received prior midostaurin or cladribine. The 79 midostaurin patients came from the German Registry on Eosinophils and Mast Cells, treated between 2009 and 2021, with 20% having prior cladribine. Both groups were dominated by SM with an associated hematologic neoplasm (68% and 63%), and roughly 95% carried the KIT D816V driver mutation that defines most AdvSM.

Three-tier risk scores separate survival by years

The stratification was steep in both cohorts. Among avapritinib-treated patients, 95 patients (55%) scored 0 points and had not reached median overall survival; 56 patients (32%) scored 1 point with a median of 45 months; and 22 patients (13%) scored 2 to 4 points with a median of 19 months. The point weights came from SETBP1 (hazard ratio 6.369, 2 points), RUNX1 (2.717, 1 point), and SRSF2 (2.457, 1 point). ASXL1 was associated with only numerically shorter survival in this cohort.

In the midostaurin cohort, 34 patients (43%) scored 0 points with a median overall survival of 69 months, 30 patients (38%) scored 1 point with 35 months, and 15 patients (19%) scored 2 to 4 points with 12 months. Here the weights were EZH2 (hazard ratio 4.3, 2 points), SRSF2 (2.0, 1 point), and ASXL1 (2.0, 1 point). The practical point is that a mutation that carries weight on one drug may not carry the same weight on the other, so a risk score built for midostaurin should not be applied to avapritinib patients.

Where an SRSF2 mutation sits matters

SRSF2 appears in both panels, but the avapritinib data show that not all SRSF2 mutations behave alike. Among 64 avapritinib-treated patients with an SRSF2 mutation, 27 (42%) had a hotspot change at codon p95 and 37 (58%) had variants downstream of p95. Median overall survival was 45 months with a p95 hotspot mutation, 52 months with a downstream variant, and not reached in patients without any SRSF2 mutation. Different p95 substitutions (P95H, P95L, P95R, P95RP, P95T) did not produce distinct survival outcomes.

This matters for reading the risk score: the score counts SRSF2 as a single binary variable, while the underlying survival curves suggest a gradient from no mutation to downstream variant to p95 hotspot. The paper does not resolve whether the score should be refined to reflect that gradient, so the three-tier system remains the validated summary while the subtype analysis explains part of its biological signal.

An early KIT D816V drop predicts avapritinib survival

For avapritinib, the paper reports a rapid early response in the blood. Median KIT D816V variant allele frequency fell by 12.8% at day 15, 42.8% at one month, and 62.8% at two months, while serum tryptase fell by 70.0%, 79.1%, and 84.5% over the same points. A reduction of at least 10% in KIT D816V variant allele frequency at day 15 was associated with a more favorable median overall survival, not reached versus 47 months.

Tryptase kinetics told a similar story. A tryptase reduction of at least 50% within 30 days, or at least 60% by month 3, was predictive of improved overall survival, again not reached versus 47 months. These analyses could not be repeated in the midostaurin cohort because of the retrospective nature of data collection, so the early KIT and tryptase thresholds are avapritinib-specific findings rather than class-wide rules.

An alkaline phosphatase flare signals better outcome

One of the least intuitive results concerns alkaline phosphatase (AP), a serum enzyme measured in standard chemistry panels. Within three months of starting treatment, 82 avapritinib patients (47%) and 36 midostaurin patients (46%) had an early rise of at least 25% in AP, termed an AP flare. In both cohorts, patients with a flare had significantly lower baseline AP values than those without: 134.5 versus 228.0 U/L on avapritinib, and 158 versus 286 U/L on midostaurin.

The flare was associated with longer survival, not shorter. Median overall survival was not reached for avapritinib patients with a flare, 99 months for midostaurin patients with a flare, versus 52 and 42 months for patients with an early AP decrease of at least 25%, and 42 and 40 months for the remaining patients. Most flaring patients later fell below baseline within six months, and 52% of avapritinib and 53% of midostaurin flaring patients normalized AP. The authors present this as a dynamic laboratory marker, not as evidence that raising AP is beneficial.

Six-month normalization carries independent weight

Multivariable analyses asked which first-six-month changes remained predictive after adjustment. For avapritinib, normalization of hemoglobin and/or platelet counts (hazard ratio 0.275) and normalization of serum albumin (0.177) were significant predictors of prolonged overall survival. For midostaurin, a KIT D816V allele burden reduction of at least 25% (0.377) and normalization of hemoglobin and/or platelet counts (0.341) were significant.

The two drugs also differed in depth of response. Avapritinib rendered KIT D816V undetectable in peripheral blood in 24% of patients, while no such conversions occurred with midostaurin. Tryptase normalized below 11.4 ng/mL in 57% versus 10% of patients, and albumin normalized in 88% versus 65%. These are descriptive comparisons between cohorts assembled from different sources, not randomized head-to-head evidence, so they should be read as differences in observed response depth rather than proven drug superiority.

What the risk tiers do not settle

The most decision-relevant reading is that baseline mutation panels and early dynamic markers answer different questions and should be used together. The mutation score tells a clinician which patient is likely to do poorly before treatment starts; the day-15 KIT D816V drop, the AP flare, and six-month albumin or blood-count normalization describe how that patient is actually responding. A patient with a high-risk baseline score who shows an early KIT D816V reduction and an AP flare is not in the same situation as one with the same score and no early response.

The boundaries are equally specific. The avapritinib cohort came from trials with dose escalation in EXPLORER and mostly 200 mg daily in PATHFINDER, while the midostaurin cohort was registry-based with 100 mg twice daily in 80% of patients. The early KIT and tryptase analyses were not possible for midostaurin. The mutation panels were derived and weighted within each cohort, so the point values are cohort-specific. Within those limits, the paper's priority is clear: pair a drug-specific baseline mutation score with early on-treatment KIT D816V and laboratory dynamics, rather than relying on any single marker.

Sources used in this answer