Invited Commentary
September 9, 2024
Charting the Waters of Sickle Cell Disease With Target Trial Emulation
A. Parker Ruhl, Matthew M. Hsieh, Elizabeth A. Stuart
JAMA Intern Med. Published online September 9, 2024. doi:10.1001/jamainternmed.2024.4435
Large administrative databases hold promise for advancing biomedical care, like sails awaiting the winds of a clinically relevant question to navigate uncharted waters. However, making best use of those data for causal questions requires careful study design and analysis. Taking the helm, Alwang and colleagues1 have performed a multicenter, retrospective cohort and target trial emulation study (which seeks to emulate a randomized clinical trial [RCT]) to evaluate the comparative effectiveness of intravenous (IV) lactated Ringer vs normal saline in hospitalized patients with sickle cell disease (SCD) during vaso-occlusive episodes (VOEs), reported in JAMA Internal Medicine.1
SCD is a monogenic hemoglobinopathy with the annual number of global births expected to increase to 400 000 by 2050.2 The world over, VOEs are the cardinal manifestation of SCD. Evaluating patients with VOE in an acute care setting or emergency department begins with soliciting medical history to identify 1 or more triggers and assessing the patient’s volume status for signs of dehydration. Common VOE triggers include dehydration, infection, abrupt environmental temperature changes, and both physiologic and psychologic stress. For years, hydroxyurea was the only disease-modifying medication, where it reactivated fetal hemoglobin production to dilute sickle hemoglobin. More medications have now been approved. L-glutamine reduces oxidative stress and may prolong survival of sickle erythrocytes. Voxelotor enhances sickle hemoglobin to bind oxygen more avidly, thereby reducing the propensity to sickle. Crizanlizumab interrupts the binding of erythrocytes to vascular endothelium through p-selectin blockade to reduce erythrocyte adherence to blood vessels. Gene therapy, where a patient’s own bone marrow stem cells are modified to include optimized beta-globin (T87Q) or to reactivate fetal hemoglobin by clustered regularly interspaced short palindromic repeats (CRISPR) editing is here. These 2 cellular therapies provide alternatives to the existing allogeneic hematopoietic cell transplant, where marrow stem cells from healthy family donors replace a patient’s own sickle stem cells. However, these costly and complex interventions will not reach everyone. High-quality clinical studies assessing widely available modalities integral to clinical care guidelines, such as IV fluid resuscitation, are urgently needed.
The motivation for the work by Alwang and colleagues1 is clear: to identify the preferred fluid type for patients with VOE requiring IV fluid resuscitation. Dehydrated sickle red cells have increased intracellular concentration of sickle hemoglobin, which has been thought to further worsen sickling and capillary vaso-occlusion. In vitro experiments have demonstrated that hypotonic fluids may improve rheological properties of sickled red blood cells.3 Previously, dextrose (5%) in hypotonic saline (0.45% NaCl) was considered to rehydrate sickled red cells. Increasingly, isotonic normal saline (0.9%) has been used as it is ubiquitous; in a multicenter RCT evaluating IV magnesium infusion to treat acute VOE in children, normal saline was used in the placebo arm.4 However, normal saline carries a known risk for hyperchloremic metabolic acidosis in all patients and may promote erythrocyte sickling in people living with SCD.5
In this large, well-designed study, the authors formulated their causal question clearly, using observational data to emulate an RCT using several core principles necessary for causal inference. First, they detailed their clinically relevant and actionable causal question through the framework of a protocol for a specific RCT; in this case, comparing lactated Ringer to normal saline during VOE hospitalization with the primary outcome of hospital-free days. Second, they delineated temporal ordering by defining treatment time clearly (day 1 of hospitalization) and used inclusion and exclusion criteria and covariates measured before the exposure of interest, with outcomes measured afterwards. Third, they selected a database with extensive confounder data.
Alwang et al1 investigated the comparative effectiveness of lactated Ringer vs normal saline IV fluid resuscitation using a target trial emulation in adult patients with VOE in the national Premiere PINC AI Healthcare database. Using targeted maximum likelihood estimation within this framework, they found that, in patients who received either lactated Ringer or normal saline on hospital day 1, those receiving lactated Ringer (n = 3495) had more hospital-free days compared to those receiving normal saline (n = 52 079), with a marginal mean difference of 0.4 (95% CI, 0.1-0.6) days. Concordantly, they found that patients who received lactated Ringer had shorter hospital stays (marginal mean difference, −0.4 [95% CI, −0.7 to −0.1] days) and lower risk of 30-day readmission (marginal risk difference, −5.8 [95% CI, −9.8 to −1.8]), along with more opioid-free days (marginal mean difference, 0.4 [95% CI, 0.1-0.6] days). Notably, there were no differences in clinical outcomes such as hospital mortality, organ support–free days, and blood transfusion–free days.
These data reveal small improvements in hospital- and opioid-free days when patients received lactated Ringer. As the authors allude to, whether these differences meet a minimal clinically important difference threshold remains to be determined. This is particularly true given no differences in mortality or several other clinically relevant outcomes was found. People living with SCD are at increased risk for kidney failure, congestive heart failure, and pulmonary hypertension, among potential factors like infection, sepsis, and acute chest syndrome during VOE, making decisions on the ideal volume of fluid resuscitation nuanced on many occasions. These data do not inform decisions regarding the amount of fluid resuscitation that should be given, and in fact, the volume status of patients in this study is unknown. In addition, this study did not evaluate the use of hypotonic fluids such as a half normal saline (0.45% NaCl). The use of hypotonic fluids deserves more study given existing evidence that they may decrease sickling. This highlights some of the challenges in using data collected for a different purpose for causal questions such as this, even with a careful design, including that neither patients nor clinicians can be masked, novel treatments or biomarkers cannot be assessed, and some confounders may be unmeasured or imprecisely measured. Each study needs to be assessed on its own merits, design, and potential residual biases.
As the field of target trial emulation matures, we need clear guidance on how to assess such nonexperimental studies. Prespecification of protocols and analysis plans for target trials, including setting a priori hypotheses, is important. The in-progress reporting checklist, Development of the Transparent Reporting of Observational Studies Emulating a Target Trial (TARGET) guideline,6 akin to CONSORT for RCTs and STROBE for observational data, is a key step.
The study by Alwang and colleagues1 should be heralded because it is an apt use of existing population-level observational data from an administrative database to support the clinical research of a well-defined and highly relevant clinical question in an understudied population. This study found small but significant improvements in hospital-free and opioid-free days with the administration of lactated Ringer in adults. Its findings direct us to redouble our scientific efforts to better elucidate the underlying mechanisms and pathophysiology of erythrocyte sickling and endothelial adhesion and how they interact to affect vaso-occlusion and vascular function. Together, these studies would give us the needed insights to define a conceptual model regarding the effect of IV fluid type on SCD pathophysiology and clinical outcomes.
The more fundamentally we understand the science of sickled erythrocytes and fluid resuscitation, the closer we can come to understanding which interventions are most effective, when, and for whom, in both low- and high-resource settings. Ultimately, patients living with SCD may have more days outside of hospital care to chart their own journeys.