RT-PCR-Based MRD Detection in NPM1 Mutated AML: A Prospective Follow-Up Study in 130 Patients.

To evaluate the applicability of NPM1 mutations as follow up markers and to evaluate whether there is any prognostic relevance for NPM1 detection in MRD we studied 531 samples of 130 NPM1 mutated AML during and after therapy. NPM1 mutations are heterogeneous with 80% belonging to type A, 10% to types B and D, and further 10% to more than 40 different yet described rare types. We have established NPM1 specific real time PCR assays for 12 different NPM1 variants. This study included 396 samples (spl) of 98 patients (pts) with type A, 35 spl of 8 pts with type B, 63 spl of 11 pts with type D and 37 spl of 13 pts with 9 different rare types. All assays are RNA based rendering high sensitivities between 1:100,000 to 1:1,000,000. Seven of these assays are specific, while in five of them a minimal detection of the wildtype allele cannot be avoided. Per each patient 2–15 samples were analyzed (mean 4.1, median: 3). Mean follow-up time was 302 days (median: 220 days). The initial expression ratio (% NPM1 mutation/ABL) varied between 21,8 and 8842 (median 540,6). No correlation of this initial ratio to clinical outcome was found. The expression ratios of the NPM1 mutations during follow up correlated well to clinical follow up. In 84 cases with continuing clinical remission (CCR) NPM1 mutation levels were either below traceability or levels were very low or still decreasing. Seven cases were resistant to chemotherapy what was also identified by non-decreasing NPM1 levels. Relapses were detected in 35 cases. At relapse all cases had high NPM1 levels comparable to those at diagnosis. Of these 35 relapses 16 were predictable by increasing NPM1 levels 31–126 days (median 73 days) before clinical relapse. Further eight cases did not reach more than a three log reduction after consolidation and relapsed within the first year after start of therapy. In 11 cases early detection of relapse by PCR was not possible due to lack of samples within the last 6 months before clinical relapse was diagnosed. Four cases had increasing levels at the last follow up with still unclear outcome. Based on these findings 5 different patterns were defined:

1. good response, CCR (n=84),

2. good response, predictable relapse (n=16),

3. < 3 log reduction, followed by relapse (n=8),

4. resistant AML (n=7),

5. non predictable relapse or unclear outcome due to missing MRD samples (n=15).

To analyse the impact of NPM1 mutation levels on prognosis four different follow-up intervals were defined: interval 1: day 21–60 after start of therapy; interval 2: days 61–120; interval 3: day 121–365, 4: later. Using Cox regression analysis a significant impact of MRD levels on EFS was detected for interval 1 (p=0.033), interval 2 (p<0.001), interval 3 (p=0.002) and interval 4 (0.037). Furtheron for interval 4 a threshold of 0.01% NPM1/ABL was defined. Thirty cases had a % NPM1/ABL of <0.01% and 28 cases >0.01%. Both groups had significantly different EFS rates (after 2 years: 96.7% vs. 61.5%; p<0.001). In conclusion

1. NPM1 mutations are highly effective and sensitive markers for PCR-based MRD detection in normal karyotype AML,

2. mutation levels during follow up highly correlate with the clinical course of the disease and are highly predictive for prognosis,

3. early detection of relapse is possible by increasing MRD levels up to four month before clinical relapse.