External Validation of the French Score in Thrombotic Microangiopathies: Real-World Experience in a Large Retrospective Cohort of 1483 Patients

INTRODUCTION

Thrombotic thrombocytopenic purpura (TTP) is a hematological emergency characterized by a deficiency in ADAMTS13 activity (<10%), with high mortality within the first 48 hours without treatment. Due to clinical similarities with other thrombotic microangiopathies (TMA) and the lack of urgent availability of ADAMTS13 activity testing, various scores are used for early suspicion. The French score is widely used in clinical practice. However, this score has been validated in a small number of cohorts with limited sample sizes.

OBJECTIVES

- To perform an external validation of the French score in a cohort of 1483 patients with TMA and to assess its performance according to international guidelines recommendations.

- To review previous external validations of the French score.

MATERIALS AND METHODS

This is a retrospective observational study (2012-2023) that included 1483 adult patients with a first episode of TMA (anemia or microangiopathic hemolysis with >1% schistocytes, acute thrombocytopenia, and elevated LDH) and with confirmation or exclusion of TTP by ADAMTS13 levels. The probability of TTP was calculated using the French score, according to the criteria of the latest update of the ISTH guidelines. This scale includes thrombocytopenia (<30x10^9/L) and the absence of renal failure (Cr <2.26 mg/dL), awarding 1 point (p) for each item. Thus, 1 point on the French score indicates an intermediate probability of TTP and 2 points a high probability. Patients with disseminated intravascular coagulation, hematopoietic stem cell or solid organ transplantation, a prior history of cancer or active cancer, or sepsis are directly classified as having a low probability of TTP (0p). The performance of the French score [sensitivity (S) and positive predictive value (PPV)] was analyzed in this cohort and in three additional studies that performed an external validation of the French score in TMA.

RESULTS

In our cohort, the prevalence of TTP was determined to be 18.2% (n=270), consistent with data described in other large cohorts of TMA patients. For intermediate probability of TTP (≥1p), a S of 92% and a PPV of 40% were obtained. For high probability evaluation (2p), the S and PPV were 78.9% and 70.2%, respectively.

Previous studies (Burguet et al., Liu et al., Fage et al.) analyzed cohorts between 124-132 patients. For intermediate probability of TTP (≥1p), sensitivities were 100% and 94.6%, respectively (data for intermediate risk were not available in the Fage et al. study). The PPV in the three studies was 70.2%, 88.9%, and 16%, respectively. For high probability classification of TTP, sensitivities were 78.9%, 83.3%, and 57%, and the PPV was 93%, 88.9%, and 16%, respectively.

CONCLUSIONS

In our cohort, the French score achieved adequate sensitivity for intermediate probability cases (≥1p). However, nearly one in ten TTP patients was a false negative, with the majority (70%) having a history of cancer. Previous studies obtained similar sensitivity results for this cut off.

For high probability evaluation (2p), 70% of patients with this score are TTP cases, which is far from the 90% PPV recommended by international guidelines to start caplacizumab before the ADAMTS13 result is available.

The studies analyzed showed variable PPV values (16-93%) for patients with high probability of TTP. In the Liu et al. and Burguet et al. studies, an adequate PPV was observed; however, this value is overestimated given the prevalence of TTP in both cohorts is too high (57 and 62,5%) due to selection bias. In the Fage et al. study, the PPV was lower than expected due to fewer TTP cases recruited, likely due to more permissive TMA criteria (only 6,2% were TTP). Considering that our cohort reflects more accurately the real prevalence of the disease, our data is more reliable in terms of PPV.

Another widely used scale is PLASMIC score. In our cohort, we have been unable to perform external validation of it due to the lack of some variables required in this score. However, different external validations with large sample sizes (n > 100) yield a positive predictive value (PPV) ranging from 12% to 76.5%. Therefore, we consider it necessary to create new scores or modify existing ones based on large, well-selected cohorts of patients to improve PPV according to the ISTH guidelines. This way, the number of patients with TMA treated with TTP-specific drugs (caplacizumab) without meeting diagnostic criteria of TTP could be reduced.