Dissecting heterogeneous outcomes for paediatric Burkitt lymphoma in Malawi after anthracycline-based treatment

We read with interest the recent paper reporting improved outcomes for paediatric Burkitt lymphoma after the addition of doxorubicin 120 mg/m² to the previously described 28-day regimen in Blantyre, Malawi (Molyneux et al, 2016), and are grateful for that group’s dedicated efforts over many years to improve care for this common childhood cancer throughout the region. The prior 28-day regimen without doxorubicin includes cyclophosphamide, prednisone and vincristine, with intrathecal methotrexate and hydrocortisone (Depani et al, 2015). With the addition of doxorubicin, Molyneux et al (2016) reported improved disease-free survival from 28% to 66% at 12 months for stage III/IV disease.

Importantly, of 84 children with Burkitt lymphoma during the study period, 26 (31%) were excluded, resulting in only 58 evaluable children. Of these 26 exclusions, six were treated on a different protocol, 10 died at presentation before the diagnosis was confirmed and chemotherapy started, three absconded during treatment and seven were lost to followup. Removing very sick children from analyses improves survival relative to descriptions of more unselected patient populations. Additionally, in sub-Saharan Africa, failure to appropriately account for loss to follow-up results in overestimated survival for human immunodeficiency virus (HIV), HIV-associated Kaposi sarcoma and paediatric lymphoma, as mortality is a major cause of loss to follow-up in these environments, where death often occurs at home and may be frequently unascertained without very active tracing (Stanley et al, 2016a). We believe these are important considerations in contextualizing otherwise important data presented by Molyneux and colleagues.

For comparison, in the capital of Malawi, Lilongwe, our group recently reported results for treatment of paediatric Burkitt lymphoma with six cycles of CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone, with intrathecal methotrexate and hydrocortisone) (Stanley et al, 2016b). Relative to the Blantyre protocol incorporating anthracyclines, our approach was somewhat more aggressive and longer in duration, with three additional doses of cyclophosphamide, vincristine and intrathecal treatment, four additional doses of doxorubicin (total dose 240–300 mg/m²) and six additional courses of prednisone. Intensification at our centre was guided by abundant experience with frequent relapses after less intensive treatment. Using this approach, 12-month overall survival in Lilongwe was 71% for stage I/II, 38% for stage III and 25% for stage IV. Our cohort was unselected and analysed as intention-to-treat among all enrolled children, without exclusion of deaths before cytotoxic treatment. Additionally, we underwent very active tracing to ascertain vital status for nearly all children with loss to follow-up in only 3 of 73 enrolled children, who had returned home to neighbouring Mozambique after seeking care in Malawi and could not be traced via any method. Similarly, in the United States in the 1980s, the Children’s Cancer Group reported that the addition of daunomycin to COMP (cyclophosphamide, vincristine, low-dose methotrexate 300 mg/m² and prednisone) for non-lymphoblastic mature B-cell non-Hodgkin lymphoma resulted in 10-year event-free survival of only 57%, compared to 55% for patients receiving COMP alone (Sposto et al, 2001). In addition to differences in patient selection, variations in stage assignment also probably complicate comparisons across centres in sub-Saharan Africa, given typical staging with relatively crude, highly operator-dependent technologies like ultrasound, as opposed to resource-rich settings where advanced imaging modalities are routinely applied (Marjerrison et al, 2012).

On central adjudication of all deaths in our cohort, 33 of 45 (73%) were from relapsed or refractory Burkitt lymphoma (Stanley et al, 2016b). Molyneux et al (2016) referred to 40% treatment-related mortality using CHOP from our published report, which is a departure from the 16% treatment-related mortality we described, accompanied by a supplemental table detailing each death in a de-identified manner for readers to judge. This is only slightly higher, as expected, than 12% treatment-related mortality reported by Molyneux et al (2016) using their protocol, and is actually likely to be somewhat overestimated, as we intentionally over-ascribed deaths to treatment when deaths occurred in children with advanced disease in close proximity to chemotherapy receipt. Furthermore, we have presented data evaluating quantitative plasma Epstein–Barr virus (EBV) DNA, a biomarker with demonstrated clinical utility in many EBV-associated malignancies, and found that 70% of children with paediatric Burkitt lymphoma continue to have detectable EBV DNA at CHOP completion in Lilongwe (Westmoreland et al, 2017). If one believes that this may be a potentially implementable point-of-care assessment for minimal residual disease in endemic EBV-associated Burkitt lymphoma, this is further evidence suggesting that the main issue for children with advanced paediatric Burkitt lymphoma in Malawi is insufficient cytotoxic intensity to eradicate disease using CHOP or analogous strategies.

Finally, due to suboptimal outcomes with CHOP, and after the original presentation of the Blantyre data in abstract form at the International Society of Paediatric Oncology annual meeting in 2015 (Molyneux et al, 2015), we attempted to apply the Blantyre 28-day protocol with doxorubicin for stage III/IV patients in Lilongwe, Malawi, under the supervision of a full-time paediatric haematologist on site. Using this approach, our experience has been less encouraging for advanced Burkitt lymphoma, with only one of 16 patients (6%) currently surviving disease-free after completing the protocol. Seven (44%) had refractory or relapsed disease, four (25%) died within the first 2 weeks of treatment, and four (25%) had a partial response to treatment. As a result, we are no longer using this protocol to treat stage III/IV patients in Lilongwe. However, we have had excellent preliminary results using the Blantyre 28-day protocol without doxorubicin for stage I/II patients, and therefore continue to use this regimen for such patients currently.

In conclusion, we appreciate the major contributions of Molyneux et al (2016) to paediatric Burkitt lymphoma care in sub-Saharan Africa, believe that sharing of heterogeneous regional experience is valuable for rigorous examination, and look forward to on-going collaborative efforts to define best approaches for this disease, which we are actively pursuing with many regional partners.