CML Treatment in Dutch Hospitals with Low Patient Volumes Is Associated with a Substandard Quality of Molecular Response Monitoring

Introduction

Chronic myeloid leukemia (CML) is a relatively rare condition with approximately 170 new patients per year in the Netherlands. Patients at risk of progression to advanced stage disease need to be identified based on their molecular BCR-ABL1response, for which detailed guidelines are available. We evaluated the quality of molecular monitoring in Dutch clinical practice.

Methods

Chronic phase CML patients with a minimum of one year available follow-up were abstracted from a real-world population based registry of patients newly diagnosed between January 2008 and April 2013 in the Netherlands (PHAROS-CML). Hospitals were categorized into three groups based on the average number of CML-treatment initiations per year: low (<1/year), intermediate (1-2/year) and high volume (>2/year). The minimum standard of care was defined as molecular monitoring of response at least three times in the first year of treatment.

Results

In 253 eligible patients, molecular response assessment within the first year after diagnosis was performed a minimum of three times in 78%, twice in 15%, once in 4% and not performed in 3% of patients. Treatment for these patients was initiated in 54 different hospitals, corresponding to an average just below 1 new CML patient per year per hospital. In univariate logistic regression analysis treatment in an academic hospital, clinical trial inclusion, hospital volume and first line second generation TKI (2GTKI) treatment proved to be significant predictors of the minimum standard of care. Only the latter two remained significant predictors in multivariate analysis.

Treatment was initiated for 153 patients (61%) in high volume hospitals (n=16), 66 patients (26%) in intermediate volume hospitals (n=18) and 34 patients (13%) in low volume hospitals (n=20). In the high volume hospitals, 90% of patients received the minimum standard of care, compared to 67% in the intermediate (p<.001) and 50% in the low volume hospitals (p<.001). Moreover, monitoring in imatinib treated patients was poorer than in patients initially treated with 2GTKI (74% vs 94%, p=.002).

Conclusion

Approximately 40% of newly diagnosed CML patients in the Netherlands are treated within centres that see less than two new patients per year. Such low treatment volumes are associated with a substandard quality of molecular response monitoring. Surprisingly, patients treated with imatinib as first line treatment are monitored less stringent, even though the rate of treatment failure is known to be higher than 2GTKI treated patients. Our observations support a potential patient benefit of centralization of CML treatment.