Diagnosing Hemophagocytic Lymphohistiocytosis with Machine Learning: A Proof of Concept

Hemophagocytic lymphohistiocytosis is a hyperinflammatory syndrome characterized by uncontrolled activation of immune cells and mediators. Two diagnostic tools are widely used in clinical practice: the HLH-2004 criteria and the Hscore. Despite their good diagnostic performance, these scores were constructed after a selection of variables based on expert consensus. We propose here a machine learning approach to build a classification model for HLH in a cohort of patients selected by glycosylated ferritin dosage in our tertiary center in Lyon, France. On a dataset of 207 adult patients with 26 variables, our model showed good overall diagnostic performances with a sensitivity of 71.4% and high specificity, and positive and negative predictive values which were 100%, 100%, and 96.9%, respectively. Although generalization is difficult on a selected population, this is the first study to date to provide a machine-learning model for HLH detection. Further studies will be required to improve the machine learning model performances with a large number of HLH cases and with appropriate controls.

A retrospective survey of patients who discontinued participation in the JPLSG HLH-2004 clinical trial

Although clinical trials have reported an improvement in the prognosis of hemophagocytic lymphohistiocytosis (HLH), current treatment outcomes are unsatisfactory, especially in severe cases. Most clinical trial patients with severe disease discontinue participation due to complications associated with HLH or treatment-related toxicity. A retrospective survey of patients who discontinued participation in the JPLSG HLH-2004 clinical trial was conducted to review the detailed course of these cases to optimize HLH treatment and supportive care. Findings in these patients were compared with those of 45 patients who completed the protocol treatment. The 3 year overall survival rate of patients who completed treatment was 86.7%, versus 50.7% for those who did not complete treatment. Incidence of serious adverse events, such as infections, coagulopathy, and posterior reversible encephalopathy syndrome, during the initial 8 weeks of treatment was much higher in patients who did not complete treatment than in patients who completed treatment. To improve overall outcomes of patients with HLH, it is important to not only optimize HLH-directed therapy but also provide appropriate supportive care.

Treatment of pediatric primary hemophagocytic lymphohistiocytosis with the HLH-94/2004 regimens and hematopoietic stem cell transplantation in China

We aimed to clarify the clinical characteristics, prognostic factors, and effectiveness of the HLH-94/2004 regimens and hematopoietic stem cell transplantation (HSCT) in pediatric patients with primary hemophagocytic lymphohistiocytosis (pHLH) in China. A retrospective analysis was performed on 38 patients with pHLH at Beijing Children's Hospital. PRF1 (34.2%) and UNC13D (31.6%) were the most common mutations in the pHLH. Thirty-eight patients were treated with the HLH-94/2004 regimens after diagnosis. Twenty-six patients (72.2%) responded to first-line treatment (complete response: 55.5%, partial response: 16.7%). The median survival time was 23 months. The overall survival (OS) rate at 3 years was 74.7%. There was no significant difference in the response rate (72% vs. 63.6%, P = 0.703) or 3-year OS (83.6% vs. 66.7%, P = 0.443) between the patients treated with the HLH-94 regimen and those treated with the HLH-2004 regimen. The incidences of all side effects in patients treated with the HLH-94 or HLH-2004 regimen were 32.0% and 18.2%, respectively (P = 0.394). Among 15 patients treated with HSCT, neither the preconditioning regimen nor the donor type affected patient prognosis (P = 0.205 and P = 0.161, respectively). The disease status (remission or nonremission) before preconditioning did not affect prognosis or the incidence of GVHD. Furthermore, a higher bilirubin level (≥ 30 μmol/L) was correlated with a poorer prognosis in pHLH patients (P = 0.026). The effectiveness rates of the HLH-94 and HLH-2004 regimens, chemotherapy, and HSCT were similar in pHLH patients. A bilirubin level ≥ 30 μmol/L might be an adverse prognostic factor in pHLH.

A comprehensive analysis of adult patients with secondary hemophagocytic lymphohistiocytosis: a prospective cohort study

Secondary hemophagocytic lymphohistiocytosis (HLH) is a rare but fatal condition with various underlying disorders in adult patients and is diagnosed based on the HLH-2004 criteria, which were established based on experience in pediatric patients. However, few studies have prospectively evaluated the treatment outcomes and diagnostic performance of HLH criteria in adult patients with secondary HLH. Thus, we performed a single-center, prospective cohort study of adult patients with suspected HLH, and we analyzed treatment outcomes of patients enrolled between 2017 and 2019 as an interim analysis ( ClinicalTrials.gov Identifier: NCT03117010). Of the 73 patients with suspected HLH, 70 patients completed the evaluation for ≥ 7 of the HLH-2004 criteria, and 55 patients were diagnosed with HLH (55/73, 75%). Although serum ferritin and fever had a sensitivity of more than 90%, both had exceptionally low specificity, whereas soluble CD25 had a sensitivity of more than 90% and specificity of 80%. Forty patients with malignancy-associated HLH had B cell (n = 19) or T- or NK-cell (n = 21) lymphoid malignancy, whereas 15 patients had non-malignant disorders. Non-malignancy-associated HLH had greater than 90% 1-year overall survival (OS) after diagnosis of HLH, whereas that for malignancy-associated HLH was less than 40%. In conclusion, our study showed promising treatment outcomes for patients enrolled in our prospective cohort study, and prospectively demonstrated the diagnostic performance of the HLH-2004 criteria in adult patients with suspected HLH. Given that lymphoma was the most common cause of HLH in adults, thorough evaluation for lymphoma should be performed in adults with suspected HLH.

Outcomes in children with hemophagocytic lymphohistiocytosis treated using HLH-2004 protocol in Japan

Recent advances in intensive chemo- and immunotherapy have contributed to the outcome of hemophagocytic lymphohistiocytosis (HLH); however, the prognosis of HLH in children differs by HLH subtype. In Japan, secondary HLH, particularly Epstein-Barr virus-associated HLH (EBV-HLH), is the most common HLH subtype. The prognosis of HLH has improved in recent years. We here conducted a prospective study of 73 patients who were treated with HLH-2004 protocol in Japan. EBV-HLH, familial HLH (FHL), and HLH of unknown etiology were seen in 41, 9, and 23 patients, respectively. Patients with resistant or relapsed disease after HLH-2004 treatment and those with FHL received hematopoietic stem cell transplantation (HSCT). The induction rate after initial therapy was 58.9%, and the 3-year overall survival (OS) rate of all patients was 73.9% and differed significantly among those with EBV-HLH, FHL, and HLH of unknown etiology. Of the 17 patients who received HSCT, the 3-year OS rates of those with and without complete resolution before HSCT were 83.3% and 54.5%, respectively. Outcomes in children with HLH who were treated with the same protocol differed among HLH subtypes. Appropriate strategy for each subtype should be established in future studies.

Posterior reversible encephalopathy syndrome in pediatric patients undergoing treatment for hemophagocytic lymphohistiocytosis: clinical outcomes and putative risk factors

Background

Hemophagocytic lymphohistiocytosis (HLH) is a rare multiorgan disease of toxic immune activation caused by the interaction of cytotoxic T cells and innate immune cells and frequently involves the central nervous system (CNS). Posterior reversible encephalopathy syndrome (PRES) might develop during treatment with the HLH-2004 protocol from the Histiocyte Society. The aims of this study were to evaluate clinical outcomes and putative risk factors for prediction of PRES related to HLH.

Methods

We reviewed the medical records of 28 patients with HLH who were treated between April 2005 and April 2012. We compared various clinical and laboratory parameters in patients without or with PRES to evaluate putative risk factors related to development of PRES.

Results

Six (21.4%) of the patients experienced PRES during treatment with the HLH-2004 protocol. Clinical and laboratory manifestations were not different compared with other conditions causing PRES. The main mechanism of PRES may be related to the HLH-2004 protocol and a high pro-inflammatory state. Most patients recovered quickly from neurologic manifestations without significant long-term sequelae. Preceding hypertension, an increase in ferritin level >50% compared with 1 week before development of PRES and hyponatremia were statistically significant factors.

Conclusion

PRES is clinically reversible and has a favorable outcome in patients with HLH. Awareness of PRES and a differential diagnosis of other causes of neurologic complications, including CNS involvement of HLH, can help avoid unnecessary treatment or delayed management. Patients with preceding hypertension, hyponatremia, and rising ferritin levels during HLH treatment should be closely monitored for PRES.