This was a prospective monocentric cohort study conducted at the CRLD in Bamako, Mali, from January 2023 to December 2024. It included a cross-sectional analysis at inclusion to identify factors associated with abnormal TCD, followed by longitudinal follow-up over 24 months of children with abnormal and conditional TCD.

All children with homozygous SS or Sβ0-thalassemia SCD, aged 2 to 15 completed years, in steady state, regularly followed at the CRLD and whose parents/guardians provided written informed consent, were included. Steady state was defined as absence of vaso-occlusive crisis (VOC), acute chest syndrome (ACS), acute infection or hospitalization within the previous 15 days and absence of blood transfusion within the previous 3 months. Steady state definition was independent of genotype and reflects clinical stability [10].

Non-inclusion criteria were: 1) documented history of clinical stroke confirmed by medical records or neuroimaging—excluded to focus on primary stroke prevention; 2) blood transfusion within 3 months prior to TCD, which can artificially lower velocities; 3) occurrence of VOC, ACS or acute infectious episode within 15 days prior to examination; 4) parental refusal. Patients with absent temporal acoustic windows making the examination uninterpretable were excluded from the final regression analysis but described in the flow-chart. Exhaustive sampling was used: all eligible children presenting for follow-up consultation during the study period were offered TCD. In total, 207 children were included.

Each child underwent a standardized clinical consultation with collection of sociodemographic data (age, sex), SCD genotype (SS, Sβ0-thalassemia), history of VOC ≥3/year, ACS, splenic sequestration, hospitalization, and current disease-modifying therapy. Physical examination was performed (Table 1).

Table 1. Sociodemographic and clinical characteristics of 207 children with sickle cell disease included.

A baseline biological assessment including complete blood count with reticulocyte count, hemoglobin electrophoresis and iron status was performed in steady state, on the same day or within 7 days preceding the index TCD (median 0 days, interquartile range [IQR] 0 - 2 days), at least 4 weeks after any acute event. Baseline hemoglobin was defined as hemoglobin level measured in steady state.

The study received approval from the Scientific Committee of the CRLD (Institutional Review Board of the CRLD, approval number: 2022/18/CE-CRLD) and was conducted in accordance with the Declaration of Helsinki. Written informed consent from parents/guardians and assent from children aged > 7 years were obtained. Data confidentiality was respected through anonymization. Therapeutic management was independent of participation in the study.

TCD was performed according to the STOP protocol, by a single trained and certified operator, using a 2 MHz pulsed Doppler device (Sonosite M-Turbo®). The examination was bilateral, exploring the middle cerebral arteries (MCA), anterior cerebral arteries (ACA) and posterior cerebral arteries (PCA) as well as the terminal internal carotid arteries (ICA) and carotid siphons. The angle-corrected TAMMV was retained for each vessel (Table 2 and Table 3).

Table 2. Initial transcranial doppler results in 207 children.

Table 3. Factors associated with abnormal transcranial Doppler in univariate and multivariate analysis (n = 204, excluding 3 uninterpretable).

Stroke risk classification was based on the highest TAMMV measured in the MCA or ICA. TCD was considered normal if TAMMV < 170 cm/s, conditional if 170 cm/s ≤ TAMMV < 200 cm/s, abnormal if TAMMV ≥ 200 cm/s, and uninterpretable if no exploitable Doppler signal was obtained despite adequate windows.

Children with abnormal TCD were placed on alternative (non-combined) disease-modifying therapy: monthly transfusion program targeting HbS < 30% as first-line if phenotyped secure blood was available and parental acceptance was obtained; otherwise, optimized hydroxyurea at 20 - 35 mg/kg/day was proposed as an alternative, in accordance with recent African trials SPIN, SPRING and SPHERE and ASH recommendations for resource-limited settings. Criteria for attribution were documented prospectively.

Adherence in the abnormal TCD group was defined a priori as: for transfusion, ≥10/12 sessions per year with HbS < 30% for at least 6 months; for hydroxyurea, intake ≥ 5 days/7 as declared, confirmed by pharmacy dispensing and increase in mean corpuscular volume (MCV) and fetal hemoglobin (HbF). Non-adherence was defined as <50% of transfusion sessions or HU interruption > 2 months.

Conditional TCDs were controlled at 3 months; high conditional (185 - 199 cm/s) was controlled at 1 month then every 3 months. All patients had TCD follow-up at 12 months.

Data were entered into REDCap and analyzed with SPSS version 25.0. Quantitative variables were expressed as mean ± standard deviation (SD) or median with IQR. Qualitative variables were expressed as numbers and percentages. Comparison of means used Student’s t-test or Mann-Whitney test. Comparison of proportions used Chi² test or Fisher’s exact test. Multivariate logistic regression was performed to identify factors independently associated with abnormal TCD. The 3 uninterpretable TCDs were excluded from regression, yielding analysis on 204 children (18 abnormal vs 186 non-abnormal: 152 normal + 34 conditional). Given 18 events, the final model was limited to 2 variables according to the rule of 1 variable per 9 events. Variables with p < 0.20 in univariate analysis (age < 8 years, Hb ≤ 7 g/dL, history of ACS, ≥3 VOC/year, absence of HU) were entered, with backward stepwise selection. Adjusted odds ratios (aOR) with 95% confidence intervals (CI) are presented. Significance threshold was set at p < 0.05.

During the study period, 312 children were screened for TCD in the CRLD consultation. Of these, 45 were excluded: 12 with documented history of clinical stroke, 15 with transfusion within 3 months, 10 with VOC/ACS/acute infection <15 days, 5 with parental refusal, and 3 with absent temporal acoustic windows rendering examination uninterpretable. An additional 60 eligible children were not included due to incomplete data or loss before consent. Thus, 207 children were included. Of 207 TCDs, 204 were interpretable (98.5%). A flow-chart is presented asFigure 1.

Figure 1. Participant flow-chart. CRLD = Center for Research and Control of Sickle Cell Disease; VOC = Vaso-occlusive crisis; ACS = Acute chest syndrome; TCD = Transcranial Doppler; STOP = Stroke Prevention Trial in Sickle Cell Anemia.

Figure 1: Flow-chart of study participants

Flow-chart of screening, exclusion, inclusion and 24-month follow-up of children with sickle cell disease undergoing transcranial Doppler (TCD) at CRLD, Mali, January 2023 - December 2024. TCD classification according to STOP criteria: normal < 170 cm/s, conditional 170 - 199 cm/s, abnormal ≥ 200 cm/s.

Mean age was 8.3 ± 3.6 years, range 2 - 15 years. Sex ratio was 1.1 (109 males, 52.7%, 98 females, 47.3%). SS phenotype represented 94.2% (n = 195) and Sβ0-thalassemia 5.8% (n = 12). Mean baseline hemoglobin (steady state, pre-TCD) was 7.8 ± 1.2 g/dL. Notable history included ACS in 22.7% (n = 47), splenic sequestration in 11.1% (n = 23), and hospitalization for VOC > 3 times/year in 31.4% (n = 65). At inclusion, 28.5% (n = 59) were on hydroxyurea. Mean TAMMV in the MCA was 148 ± 32 cm/s.

Distribution according to STOP classification (n = 207): normal TCD 152 (73.4%, 95% CI 67.0 - 79.2), conditional 34 (16.4%, 11.7 - 22.1), abnormal 18 (8.7%, 5.2 - 13.4), uninterpretable 3 (1.5%, 0.3 - 4.2). In the 18 patients with abnormal TCD, mean TAMMV was 221 ± 18 cm/s.

In univariate analysis (n = 204, excluding uninterpretable), factors significantly associated with abnormal TCD were age < 8 years (OR 2.8, 95% CI 1.1 - 7.3, p = 0.02), baseline Hb ≤ 7 g/dL (OR 3.1, 95% CI 1.2 - 8.1, p = 0.01), history of ACS (OR 2.9, 95% CI 1.1 - 7.8, p = 0.03), and absence of hydroxyurea treatment (OR 4.2, 95% CI 1.0 - 18.5, p = 0.04). In multivariate analysis, only age < 8 years (aOR 2.6, 95% CI 1.0 - 6.9, p = 0.04) and baseline Hb ≤ 7 g/dL (aOR 2.9, 95% CI 1.1 - 7.7, p = 0.03) remained independently associated. History of ACS showed a trend (aOR 2.9, 95% CI 0.9 - 9.3, p = 0.07).

All 18 patients with abnormal TCD were treated: 11 with chronic transfusion program and 7 with hydroxyurea. After 24 months, 12 patients (66.7%) normalized their TCD (TAMMV < 170 cm/s), 4 (22.2%) remained abnormal, and 2 (11.1%) were lost to follow-up. Two patients met criteria for non-adherence. No clinical stroke occurred in the 13 adherent patients with initially abnormal TCD.

Among 34 conditional TCDs, 21 (61.8%) normalized, 9 (26.5%) remained stable, and 4 (11.7%) progressed to abnormal TCD requiring therapy.

Overall incidence of clinical stroke over 2 years was 0.96% (2 cases/207). Both strokes occurred in non-adherent children with abnormal TCD. No stroke occurred in children with normal or stabilized conditional TCD.

Our study found 8.7% abnormal TCD in children with SCD without prior stroke. This rate is comparable to 9.4% reported by Lagunju et al. in Ibadan, Nigeria over 5 years [11], but lower than 12.2% in the French newborn cohort of Bernaudin et al. [12] followed into adulthood. This difference is explained by our exclusion of history of clinical stroke (primary prevention cohort) and younger population (2 - 15 years, mean 8.3 years) compared to Bernaudin et al. who included longer follow-up. The reported prevalence of high stroke risk (abnormal TCD) in sub-Saharan Africa varies from 2.9% to 16.9% depending on study design, age range and inclusion criteria, as summarized in the systematic review by Kassim et al. (Blood Rev 2024) which included 12 studies from 2000 - 2023 (n = 4521 children) [13] and the multicenter study by Noubiap et al. 2022 (n = 1145 across 6 Nigerian centers) [14]; our result falls within this range and is consistent with regional data.

The rate of 16.4% conditional TCD is consistent with 15% - 20% in the literature [6]. The progression of 11.7% of conditional TCD to abnormal confirms the need for close monitoring every 3 months as recommended by ASH 2020 guidelines and Brazilian guidelines 2021: control at 1 month if high conditional (185 - 199 cm/s), then every 3 months [3][6].

The association between severe anemia (Hb ≤ 7 g/dL) and abnormal TCD corroborates our previous work in Mali [9] and recent studies showing that anemia increases compensatory cerebral blood flow velocity to maintain cerebral oxygen delivery [15]. The inverse relationship between hemoglobin and cerebral blood flow velocity is well established: lower Hb reduces arterial oxygen content, leading to compensatory cerebral vasodilation and increased velocities.

Young age < 8 years as a risk factor is classic, with vasculopathy beginning in the first years of life, justifying screening from 2 years of age [3][12]. The Blood Reviews 2024 meta-analysis confirms that early screening significantly reduces mortality and neurocognitive sequelae [13].

Our work demonstrates feasibility and effectiveness of TCD screening and management in routine practice in a resource-limited African setting: 66.7% normalization of abnormal TCD and stroke incidence of 0.96% over 2 years. These results are close to STOP trial where stroke risk decreased from 10% per year without treatment to ~1% per year under chronic transfusion [4], and to the meta-analysis by Kassim et al. showing stroke rate of 1.0 per 100 patient-years under hydroxyurea or transfusion vs 10.7 without treatment [13].

Hydroxyurea was used in 7 patients due to lack of sustainable access to secure phenotyped blood. Recent African trials are changing the paradigm: SPIN trial (Nigeria, 2020) showed that moderate fixed dose 20 mg/kg/day lowers TAMMV to <200 cm/s within 3 months in majority of children with abnormal TCD [16]. SPRING trial, phase 3 multicenter randomized double-blind trial (2022), confirmed efficacy of moderate-dose hydroxyurea in primary stroke prevention in resource-limited settings [17]. SPHERE trial in Tanzania (2024) with dose escalation showed sustained reduction in velocities [18]. REACH trial (4 African countries, 3-year follow-up 2022) confirms tolerability and mean decrease of 20 cm/s in TCD velocities under hydroxyurea [19]. Long-term TWiTCH trial (2020) shows that after 4 years of transfusion without severe vasculopathy on magnetic resonance angiography (MRA), hydroxyurea can replace transfusion to maintain velocities [20]. Our finding of 5/7 patients on hydroxyurea normalizing TCD is consistent with these data and encouraging for resource-limited countries [16]-[19].

The 2 strokes occurring in non-adherent patients highlight importance of therapeutic education and adherence support, a major barrier identified in DISPLACE study in USA and Africa [7].

Our study is monocentric, limiting generalizability. Absence of magnetic resonance imaging (MRI) for silent infarcts and MRA for vasculopathy grading remains a limitation, as highlighted by WHO AFRO 2024 which advocates for introduction of TCD and hydroxyurea into minimum package of health services [7]. Small number of abnormal TCD events (n = 18) limits power of multivariate model, as acknowledged, and precludes definitive conclusions on stroke prevention efficacy. Loss to follow-up of 2 patients with abnormal TCD may have underestimated adverse outcomes. Finally, adherence assessment was partly based on declaration and pharmacy records, without electronic monitoring.