Pharmacotherapeutical strategies in the prevention of acute, vaso-occlusive pain in sickle cell disease: a systematic review

Sickle-cell disease (SCD) is a chronic, hereditary hemoglobinopathy. Every year, more than 300 000 children are born with this disease, making it 1 of the most common genetic disorders in the world. This number is expected to increase even further in the coming decades.^1,2  A majority of these patients are located on the African continent.

SCD is characterized by chronic anemia, progressive organ damage, and a wide spectrum of acute manifestations such as stroke, acute chest syndrome, and serious bacterial infections.³  As a result of these severe complications, SCD has been associated with a significant reduction in life expectancy.^4,5  One of the most frequent and debilitating symptoms of the disease is the incidence of acute, painful vaso-occlusive crises (VOCs), which typically occur in the extremities, back, joints, abdomen, or chest.^6-8  These severe and unpredictable painful attacks are responsible for >90% of acute hospital admissions in SCD.⁹  Moreover, patients frequently experience pain at home without seeking any health care support.^10,11  Aside from the direct physical morbidity of these recurrent painful attacks, other significant outcomes such as quality of life, school attendance, and societal participation of patients are greatly affected.^12-18  Moreover, high societal costs are involved. Direct costs arise from high health care utilization, and indirect costs include absence from work associated with impaired physical ability to perform daily tasks.¹⁹ 

Over the past few decades, our understanding of the pathophysiology of VOCs in SCD has increased greatly. In addition to the fundamental concept of sickle hemoglobin (HbS) polymerization upon deoxygenation and the subsequent sickling of red blood cells, the roles of other significant processes in the pathophysiology of VOCs have been identified. Adhesive interactions between leukocytes, red blood cells, and an activated endothelium add to vascular obstruction, triggered by factors such as inflammation, stress, and hemolysis.²⁰  Recurrent episodes of ischemia and subsequent reperfusion stimulate the release of reactive oxidants, thereby further promoting tissue damage, inflammation, and cellular activation.²⁰  Secondary to these processes, activation of coagulation occurs, which amplifies vascular obstruction.²¹  Thus, the pathophysiology of VOCs is characterized by a complex and highly dynamic interplay of various determinants.

To relieve the clinical burden of SCD, an important research focus lies in the prevention of VOCs. The currently available pharmacotherapeutical options for prevention of VOCs are still extremely limited. The only drug with proven efficacy registered for use in SCD is hydroxyurea (HU), a potent inducer of fetal hemoglobin. HU has been shown to be highly efficacious in SCD.^22-24  However, patients receiving HU often do not experience complete relief from their symptoms, and it is currently the only disease-modifying therapy for SCD. Therefore, there remains an urgent need for new therapeutic approaches. Novel treatments based on recently identified pathophysiological mechanisms have emerged over the past years. Many are currently under investigation in preclinical or early-phase trials, as extensively discussed in other reviews.^25-27  However, some treatments have already been evaluated in phase 2 or 3 randomized clinical trials (RCTs), but a clear overview of these studies and their findings does not exist. The objective of this study was to systematically review the currently available controlled clinical trials, investigating new pharmacotherapeutical strategies in the prevention of VOCs in SCD. Hereby, we aim to facilitate the identification of promising treatments evaluated over the years and advance their clinical implementation.

We performed a systematic review following PRISMA guidelines.²⁸ 

We identified relevant trials from electronic searches in MEDLINE (Ovid MEDLINE In-Process & Other Non-Indexed Citations and Ovid MEDLINE 1946 to present), Embase (Ovid Embase Classic and Embase 1947 to present), and the Cochrane Central Register of Controlled Trials (CENTRAL). Both full-text publications as well as conference abstracts were eligible for inclusion in our study. The search in MEDLINE was performed using the terms described in supplemental Table 1. Search strategies for all other databases were modeled on the MEDLINE version. References of eligible trials and other relevant systematic reviews were screened for additional relevant studies. There were no restrictions in our search on language, publication year, or publication status to limit potential bias. Date of the last search was 30 January 2017.

Both controlled trials as well as RCTs with a concurrent control group were eligible for this review. Studies were included if they concerned patients of all ages with either homozygous or compound heterozygous SCD and evaluated prophylactic pharmacotherapeutical interventions targeting the reduction of acute VOC pain in SCD. Studies of HU were excluded from this review because these have been discussed extensively in previous reviews.^22-24,29  Both inactive control as well as active control interventions were considered.

Primary outcomes of this review were the number and/or duration of SCD-related pain days, VOCs, or hospital admissions for VOCs. Secondary outcomes included time to first VOC or hospital admission for a VOC, intensity of pain (as expressed in pain scores), quality of life, and adverse events.

Two reviewers independently screened all titles and abstracts identified in our search using the Covidence screening tool. The full text of all potentially relevant studies was obtained. Again, 2 authors independently assessed the text for final eligibility for inclusion in this review.

A standardized and piloted electronic data extraction sheet was developed based on the checklist described in the Cochrane Handbook for Systematic Reviews of Interventions (chapter 7.3).³⁰  In summary, data were collected on the study (eg, date and location of study), population, intervention, control, and defined outcomes. In addition, data on all results within the scope of this review were collected. A data extraction sheet template is included in supplemental Appendix II. Data extraction was performed by 1 author using predefined data fields and checked for accuracy by a second author.

Two authors independently assessed the methodological quality of the studies using the Cochrane collaboration’s risk-of-bias tool (chapter 8.5).³⁰  In summary, the following 6 criteria were used: randomization sequence generation, concealment of treatment allocation, blinding of participants, clinicians and outcome assessors to treatment allocation, completeness of the outcome data, and selective outcome reporting. Each criterion received a score of high, unclear, or low risk of bias. In the main results, we summarize these 6 criteria as a single risk-of-bias score from 0 to 6. The higher this score, the more items were marked with a low risk of bias. Any other potential bias is reported separately, such as uncorrected baseline imbalances, industry funding, potential conflicts of interest of the authors, or potential carryover effect in crossover trials.

Any disagreements between reviewers concerning study selection, data extraction, or quality assessment were resolved by discussion or consultation with a third reviewer.

All included studies were grouped according to the pathophysiological mechanism of action and are presented in table format. Relative effect measures were calculated and reported if the required data were available. For dichotomous outcomes, odds ratios with their 95% confidence intervals were calculated. For continuous variables, mean differences with corresponding confidence intervals were calculated. If these could not be calculated, results are shown as reported in the study. Because of the high clinical diversity and the small number of studies per intervention, all findings are presented solely narratively; no meta-analysis was performed.

All sponsors were public or nonprofit organizations that support science in general. They had no role in the design or execution of this review, the writing of the manuscript, or the decision to submit the paper for publication.

Our database searches identified a total of 1717 unique records for initial abstract screening, of which 141 were selected for full-text screening (Figure 1). Ninety-seven studies were excluded for various reasons, as noted in Figure 1 (see supplemental Appendix III for a full list of the excluded studies). Five of the included records involved conference abstracts of studies that were published separately as full papers (duplicates).^31-35  Six included records involved 3 distinct studies, all published twice as separate, full-paper publications.^36-41  The remaining total of 36 unique studies was included in this review, covering 26 different prophylactic interventions.

In the further presentation of our results, we have classified treatments per main mechanism of action: anticoagulation, antioxidants, antiadhesion, antisickling, and other interventions. Treatments with multiple pathophysiological targets were classified by their primary mechanism of action. The included trials were published between 1968 and 2016, of which 15 trials were published before the year 2000^37,42-55  and 8 in the last 5 years.^56-63  Three studies did not use a placebo intervention for their control group.^43,54,56  A majority of studies were performed in North America (n = 10),^{33,46,47,51,61,62,64-68}  Africa (n = 9),^{38,42,52,55,56,59,69-71}  and Latin America (n = 8).^{40,44,48,49,53,54,57,72}  Four studies were performed on multiple continents.^58,60,63,73  The number of patients per study ranged from 9 to 341. A majority of studies were limited to patients with the HbSS or HbSβ⁰ genotype (n = 26).^{37,38,42,44,45,48-57,59-61,63,64,66,67,69-71,74} 

A summary overview of the characteristics, risk of bias, and primary outcomes of the included studies is provided in Table 1. The complete results of our risk-of-bias assessment and a detailed overview of the primary and secondary outcomes of this review per study are provided in supplemental Tables 2 and 3, respectively.

Overall, the reporting of the included trials on methodological practices was poor. Only 7 of the 35 studies scored an overall low risk of bias on all items in the quality assessment.^{58-60,66,69,72,73}  Fourteen studies scored a high risk of bias on at least 1 of the items of the quality assessment, in particular on incomplete outcome data (n = 8) and blinding of participants (n = 6) and personnel for intervention (n = 6).^{37,43-45,48,49,51,54,56,63,64,70,71,74} 

A total of 7 studies assessed interventions targeting coagulation activation. In 2 crossover studies in children and adults (N = 49 and N = 29) performed in the 1980s, oral treatment with the antiplatelet agent aspirin (acetylsalicylic acid) did not provide benefit over placebo in the outcomes of this review.^49,51 

Cabannes et al⁵⁵  evaluated the clinical efficacy of the antiplatelet drug ticlopidine (adenosine diphosphate receptor inhibitor) in adolescent and adult patients with HbSS disease (N = 140). The authors reported a significant decrease in the total number and duration of VOCs and a nonvalidated VOC severity index (Table 1; supplemental Table 3).

In an early study in patients with HbSS disease in Brazil (N = 57), the platelet inhibitor pentoxifylline significantly reduced the number and duration of pain events and the number of patients with pain events as compared with placebo.⁴⁴  The study scored an unclear or high risk of bias for multiple items in the quality assessment. Schnog et al⁷²  evaluated the effects of the oral vitamin K antagonist acenocoumarol in a pilot crossover study (N = 22). No differences in the frequency of vaso-occlusive events were observed between the acenocoumarol and placebo arms.

The effects of the platelet inhibitor prasugrel were evaluated in both a small phase 2 study in adult patients (N = 62)⁶²  and a large multicenter phase 3 study in pediatric patients (N = 341).⁶⁰  In both studies, oral treatment with prasugrel did not provide benefit over placebo in the outcomes of this review.

Five studies assessed treatments with antioxidants. Two studies evaluated the effects of daily ω-3 fatty acid supplementation. In a pilot study by Tomer et al⁶⁵  (N = 10) in the United States, a significantly lower frequency of pain episodes was found. From 2009 to 2010, Daak et al⁵⁹  performed a large single-center RCT in Sudan among patients with HbSS disease (N = 140). The annual VOC rate, number of hospitalization days for VOC rate, and rate of hospitalizations were significantly lower in the ω-3 group as compared with placebo (Table 1). Both studies scored an overall low risk of bias.

A small pilot study by Pace et al⁶⁴  (N = 21) assessed the effect of the oral antioxidant N-acetylcysteine in SCD. The mean number of VOC episodes per person-days seemed lower in the 2 N-acetylcysteine arms with the highest doses (1200 and 2400 mg per day) as compared with the 600-mg intervention arm and the placebo arm. However, this study did not provide sufficient data for post hoc statistical comparison between treatments arms.

Daily supplementation with vitamins C and E did not have any beneficial effect on the incidence of acute SCD complications or quality of life as compared with placebo in a population of Brazilian patients with HbSS and HbSβ⁰ disease (N = 83).⁵⁷ 

Niihara et al^34,61  (N = 230) reported that treatment with the oral amino acid l-glutamine over 48 weeks in patients with HbSS and HbSβ⁰ disease reduced the number of sickle-cell crises, hospitalizations, and hospital days when compared with placebo (Table 1). Also, the number of days to first sickle-cell crisis was significantly higher and crisis severity seemed to be lower in the l-glutamine group. However, only conference abstracts of this study are currently available.

Ataga et al⁵⁸  performed an international trial assessing the effects of the monthly IV administration of high-dose and low-dose crizanlizumab (N = 198), a monoclonal antibody binding the adhesion molecule P-selectin.⁵⁸  In this study, high-dose crizanlizumab compared with placebo resulted in a decrease in the annual rate of VOCs and uncomplicated VOCs and a longer time until first VOC (Table 1; supplemental Table 3). There was no significant reduction in the annual rate of hospital admission days. The study scored a low risk of bias.

A majority of included studies assessed interventions with antisickling properties (n = 16). Three crossover studies, evaluating the effects of oral sodium bicarbonate (N = 18), promazine hydrochloride (N = 14), and cromolyn sodium nasal spray (N = 17), respectively, did not provide any beneficial effect over placebo.^43,46,74  Oral administration of the organic compound urea in an RCT in Ghana (N = 79)⁵²  and of magnesium pidolate in an RCT in the United States (N = 44; with or without HU)⁶⁸  did not show any differences between treatments arms in the outcomes of this review either.

Three studies assessed the effects of sex steroid hormones on pain in SCD. Two of these studies did not provide adequate statistical comparisons between the intervention groups (N = 44⁴²  and N = 43⁵⁴ ). De Ceulaer et al⁵³  (N = 23) did demonstrate a trend of reduction in the number of patients with bone pain episodes. However, all 3 studies scored an unclear or high risk of bias for multiple items in the quality assessment.

The clinical effects of the oral antisickling agent piracetam were assessed in 3 studies in children. The first was an early crossover study in Lebanon (N = 9) suggesting a reduction in the number of pain crises per patient per month in the piracetam group.⁴⁵  The second study (N = 87), performed in Saudi Arabia, demonstrated a significant decrease in the mean number of painful crises and hospitalizations per year and in a composite nonvalidated severity index (Table 1; supplemental Table 3).^36,37  The third study was a crossover trial in Brazil (N = 73) and did not show any benefit.^40,41  All 3 studies scored an unclear or high risk of bias for multiple items in the quality assessment.

The efficacy of 3 different phytomedicines was assessed in a randomized controlled setting, all in Nigeria. In a crossover study, Wambebe et al^38,39  compared the effects of the oral herbal preparation Niprisan with those of placebo (N = 70). The authors reported a significant decrease in the number of episodes of severe pain. However, differences in other outcomes relevant to this review did not reach significance. Akinsulie et al⁷¹  assessed the effect of the oral antisickling agent Ciklavit in children with HbSS disease (N = 100). Despite the conclusions of the authors, post hoc comparison did not show a significant difference in the number of VOCs during the study between the 2 intervention arms. Adegoke et al⁵⁶  compared the effects of oral lime juice, rich in the presumed antisickling agents ascorbic acid and phenylalanine, with those of routine care in children with HbSS disease (N = 113). A significant decrease was observed in the frequency and the number of patients with VOCs and hospitalizations. All 3 studies scored an unclear or high risk of bias for various items in the quality assessment.

Lastly, Ataga et al^66,73  performed both a phase 2 (N = 90) and phase 3 (N = 289) multicenter trial, each assessing the effects of oral treatment with the gardos channel blocker senicapoc. Both studies showed no improvement in the rate of VOCs as compared with placebo. In fact, senicapoc even seemed to increase the number of VOCs in our post hoc analysis of the latter trial. Similarly, an international multicenter trial of the oral fetal globin inducer HQK-1001 was stopped early (N = 76) because of lack of effect.⁶³ 

Four studies assessed the therapeutic value of dietary supplements. Rabb et al⁴⁸  assessed the effect of folic acid supplementation on VOCs and dactylitis in children with HbSS disease in Jamaica (N = 115).⁴⁸  In this non-RCT, there was no effect of folate on the proportion of patients with VOCs.

Three trials involved oral zinc supplementation, based on the observation that zinc may have an immunoregulatory and anti-inflammatory effect in SCD.^75-77  A large, placebo-controlled study in India (N = 130) demonstrated a significant decrease in the number of sickle-cell–related crises (including nonpain events) in the zinc sulfate group, but no effect on length of hospital stay (Table 1; supplemental Table 3).⁵⁰  Prasad et al⁴⁷  compared zinc acetate supplementation with routine care and placebo and did not observe any differences between treatments arms (N = 32). Data from an additional study by this group did not allow post hoc comparison of the number of VOCs between the treatment arms (N = 36).⁶⁷ 

Lastly, we identified 2 studies evaluating the effect of antimalarial therapies on the occurrence of VOCs in patients with HbSS disease in Nigeria and Senegal (N = 97 and N = 60), respectively. Malaria infections are an important trigger for VOCs in SCD. In the open-label, placebo-controlled study by Eke et al,⁷⁰  a significantly lower proportion of patients with VOCs was observed in the proguanil arm, along with a lower proportion of patients with hospitalizations in the pyrimethamine arm as compared with placebo (Table 1). In the study by Diop et al,⁶⁹  monthly sulfadoxine-pyrimethamine treatment as compared with placebo during high-transmission malaria season did not affect the frequency of VOCs or hospitalizations.

In this systematic review, we have formally examined the range of pharmacotherapeutical strategies beyond HU that have been evaluated over the years for the prevention of VOCs in SCD in a controlled setting. This is the first study providing a complete overview by a systematic approach, including a quality assessment.

Despite our broad search, only 36 unique studies were identified, covering 26 different interventions. A significant proportion of the included studies were relatively older and of poor quality. Eight trials were published in the past 5 years. The most promising interventions with efficacy on the frequency of VOCs or hospitalizations in SCD were the antioxidants l-glutamine and ω-3 fatty acids and the antiadhesive P-selectin inhibitor crizanlizumab.^58,59,61,65  Also, studies on the antiplatelet agent ticlopidine, the dietary supplement zinc, and antimalarial prophylaxis for patients in malaria-endemic regions provided encouraging results.^50,55,70  However, 23 of the 36 studies included in this review did not show any beneficial effect of the intervention under investigation.^{38,40,43,46-49,51-54,57,60,62,63,66-69,71-74}  The other studies were either too small or methodologically inadequate to draw strong conclusions.^{37,42,44,45,56,64} 

The amino acid l-glutamine seemed to be of great efficacy in patients with HbSS or HbSβ⁰ disease on the relevant outcomes of this review.⁶¹  The pathophysiological rationale for its effect in SCD has not yet been extensively studied. It is based on the observation that supplementation with l-glutamine enhances the nicotinamide adenine dinucleotide redox potential and may thereby exert an antioxidative effect in SCD.⁷⁸  Currently, the results of this phase 3 trial have only been presented in conference abstracts, limiting reliable assessment of the quality and results of the study and further clinical implementation.

A beneficial effect of ω-3 fatty acid supplementation in SCD was observed in 2 trials. A pilot study by Tomer et al⁶⁵  showed a significant decrease in the frequency of painful episodes compared with placebo. These findings were confirmed by a large, methodologically sound phase 3 trial by Daak et al⁵⁹  demonstrating a significant decrease in the annual VOC rate, hospitalization rate for VOCs, and number of days admitted in patients with HbSS in Sudan. These results are highly promising and are supported by multiple preclinical studies. Interestingly, ω-3 fatty acids are considered to be a multitarget intervention; in addition to having an antioxidative effect, they have been shown to inhibit platelet activation and blood-cell aggregation and inflammation and improve cell-membrane deformability.^65,79-82  However, further confirmation of its efficacy on clinical outcomes is needed. The fact that 2 new phase 3 multicenter studies are expected to start enrollment soon therefore offers hope (registered at www.clinicaltrials.gov as #NCT02604368 and #NCT02525107).

A high-dose regimen of the antiadhesive agent crizanlizumab was demonstrated to be an effective therapy in reducing the annual rate of VOCs and prolonging the time to first VOC.⁵⁸  Unfortunately, no significant reduction in the rate of admission days was reached. Possibly, this intervention is most effective in milder episodes of VOC. Also, the rate of admission days is a multifactorial outcome that is not solely determined by the severity of a VOC. Crizanlizumab is a monoclonal antibody binding the adhesion molecule P-selectin that mediates adhesion of sickle erythrocytes to the vascular endothelium.^83,84  This phase 2 study provides proof of concept that antiadhesive therapy can reduce the incidence of VOCs. Despite these encouraging results, the monthly IV administration of the drug and the likely high costs involved may prove to be barriers to widespread clinical implementation.

Oral treatment with the antiplatelet drug ticlopidine resulted in a significant reduction in the number and duration of crises and a nonvalidated severity index of crises.⁵⁵  However, outcomes were not clearly defined, baseline imbalances were not properly assessed, and it was unclear how missing data and dropouts were handled in this trial from 1984. The study by Gupta et al⁵⁰  on daily zinc sulfate supplementation in patients with HbSS disease showed a significant reduction in the number of sickle-related events.⁵⁰  However, nonpain events were also included in this definition, there were no data on zinc status at baseline, and there was an unclear risk of bias for 2 items in the quality assessment. Further research is required to assess the effect of these interventions on SCD. Eke et al⁷⁰  provided proof of concept for the efficacy of malarial prophylaxis in the reduction of VOCs and hospitalizations in patients with HbSS disease. However, these findings need confirmation in a double-blind trial design. Moreover, this intervention is only relevant to patients with SCD living in malaria-endemic regions of the world.

Our review has some limitations. Firstly, the range of pharmacotherapeutical interventions among the included studies varied significantly, hampering data synthesis by meta-analysis. Therefore, the focus of this review was to provide a broad overview, describing the quality and applicability of the included studies. Secondly, a significant portion of the included studies was of suboptimal methodological quality, limiting the reliable interpretation of these results. Lastly, by including only controlled trials, we may have missed other promising interventions that are still in earlier phases of development. However, preclinical developments and early trials have already been extensively discussed in other reviews.^25-27  In this review, we aimed to focus on interventions in advanced stages of evaluation that are closer to clinical implementation.

This comprehensive review has identified various promising pharmacotherapeutical strategies in the prevention of VOCs in SCD. Sadly, this study highlights the discrepancy between the significant burden of this disease worldwide and the low number of well-designed and adequately powered trials published so far. In future trials, we should draw lessons from prior studies and focus phase 2 to 3 trial efforts on interventions with the highest potential and a solid preclinical evidence base.⁸⁵  In addition, international consensus on a standardized set of outcomes is imperative. In conclusion, there remains a significant unmet need for new, accessible therapies for SCD. Encouragingly, the number of clinical trials in SCD has increased recently, and many novel therapies are currently under investigation.^25-27 

The authors acknowledge Arnold Leenders and René Spijker, medical information specialists at the medical library of the Academic Medical Center, for their assistance in the systematic search for this review and Lena Václavů for her critical editing of the manuscript for grammar and style.

J.W.R.S. was supported by grants from The Netherlands Organisation for Health Research and Development, the Academic Medical Center, JANIVO Stichting, Egbers Stichting, and Fonds NutsOhra, that have been awarded to K.F. and B.J.B.

Contribution: J.W.R.S. performed the search and wrote the manuscript; J.W.R.S., D.J.M., and S.C.A.T.D. performed abstract and full-text screening; J.W.R.S. and D.J.M. performed data extraction; J.W.R.S., B.J.B. and K.F. designed the study, performed quality assessments, and interpreted the data; B.J.B. and K.F. supervised the study and edited the manuscript; and all authors revised and approved the final manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Joep W. R. Sins, Academic Medical Center, Department of Hematology and Pediatric Hematology, Room H7-227, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands; e-mail: j.w.sins@amc.nl.