High Risk of Plasmodium vivax Malaria Following Splenectomy in Papua, Indonesia

In Papua, splenectomized individuals have greater risk of malaria in the 12 months following splenectomy but not of mortality. Malaria risk was higher for Plasmodium vivax than P. falciparum. Early radical cure and prophylaxis are warranted in malaria endemic areas.

The spleen plays a vital role in host immunity and erythrocyte regulation. Reduced splenic function increases the risk of severe and fatal infections, particularly those from encapsulated bacteria, and parasites such as Plasmodium and Babesia [1][2][3][4][5]. Although patients undergoing splenectomy should be vaccinated and take antimicrobial prophylaxis to minimize these risks, these are rarely delivered in resource-poor settings.
Malaria remains the most important human parasitic disease, with approximately 3.4 billion people at risk [6]. The spleen is vital for immunity to Plasmodium species [1,[7][8][9][10][11]. Splenectomized patients are more susceptible to severe falciparum malaria and have a greater risk of hospitalization and mortality [12][13][14]. Studies of splenectomized individuals in malaria-endemic regions of Malawi and Papua New Guinea suggest a risk of malaria almost double that of nonsplenectomized controls [1,2]. No studies have quantified the risk of vivax malaria after splenectomy. This is important given the recent recognition of P. vivax as a major cause of morbidity in malaria-endemic areas, including anemia and a risk of severe and fatal disease [15][16][17].
The aim of this study was to quantify the clinical consequences of splenectomy, particularly the species-specific risks of malaria, in individuals living in a malaria-endemic area in Indonesia.

METHODS
Between April 2004 and December 2013, data were prospectively collected as part of routine surveillance of patients presenting to a referral hospital in Timika, Papua, Indonesia. A retrospective analysis of this data set quantified the comparative risk of malaria and other clinical outcomes in patients who had undergone splenectomy compared to non splenectomy patients. In addition, from 2015-2016, a group of splenectomized individuals were followed prospectively and the risks of malaria over a 14-month period quantified.

Study Site
Timika is located in Mimika district, south-central Papua, Indonesia [17,18]. Approximately 200 000 people populate the area comprising highland-Papuans, lowland-Papuans and non-Papuan migrants [18]. Malaria transmission in Timika is high but unstable in lowland areas, with parasite prevalences of 13.9% and 38% by microscopy and polymerase chain reaction, respectively [19].

Hospital and Malaria Treatment
The study was conducted at Mitra Masyarakat hospital (RSMM), which served 100% of malaria patients attending surrounding healthcare facilities before November 2008, and ~80% thereafter. The hospital has 110 inpatient beds, surgical theatres, and outpatient clinics. Approximately 1800 patients are reviewed each week. The hospital laboratory does not have bacteriology facilities.
Quinine and chloroquine were used as the first-line treatment of uncomplicated malaria due to any Plasmodium species until March 2006. Following this, treatment of uncomplicated malaria changed to dihydroartemisinin-piperaquine plus a 14-day course of primaquine for vivax malaria [20]. Post-splenectomy antimalarial and antibiotic prophylaxis was not hospital policy.

Retrospective Data Collection
All patients are given a unique hospital record number (HRN) at their first hospital presentation, and all subsequent presentations can be linked to this HRN. Hospital clerks electronically record basic demographic information, mortality data, and the diagnoses given by the attending doctor (classified using the International Classification of Diseases) for each patientpresentation to hospital. Laboratory and pharmacy records are collected separately, with patients also identified by their HRN.
All inpatients and febrile outpatients are required to have a Giemsa-stained blood film checked for malaria by trained microscopists or a rapid-diagnostic-test. Previous qualitycontrol of RSMM microscopy suggests an accuracy of malaria diagnosis of >90% [17].

Identifying Splenectomy Patients
To identify all splenectomy patients accurately during the study period, surgery logbooks and registers from 2004 to 2013 were analyzed by 2 independent researchers (S. K. and B. A.) to identify individuals who may have had splenic surgery. Medical records of these individuals were then checked manually to confirm the diagnosis and reasons for surgery.

Data Preparation
Clinical and laboratory data were merged into a single database using the HRN and date of presentation. All splenectomy patients were aged between 13 and 51 years, and none were pregnant at the time of splenectomy. The primary comparative group was therefore nonsplenectomy controls, identified from nonpregnant patients aged between 12 and 60 years on their first hospital admission for any reason other than splenic surgery. In a sensitivity analysis, clinical outcomes post-splenectomy were compared to outcomes in patients who were admitted for trauma who did not undergo splenectomy (search terms for nonsplenectomy trauma control in Supplementary File 1).
Malaria reinfection or relapse within 2 weeks of an initial infection is highly unlikely [21], and therefore multiple malaria presentations within 14 days were considered to represent a single episode. No splenectomy patients were infected or represented with P. malariae or P. ovale.

Statistical Analyses
All statistical analyses were performed in SPSS v23. Baseline characteristics were presented for the splenectomized and nonsplenectomized patients and compared using the χ 2 test. In the primary analysis, the risks of representation with any malaria, P. falciparum malaria and P. vivax malaria by 12 months were compared in splenectomized versus nonsplenectomized patients. Patients were assumed to have been present in Timika for the full 12 months from the date of their splenectomy or first hospital admission (for controls). Study follow-up was curtailed at 31 December 2013. Other outcomes compared included risk of admission with other clinical outcomes (pneumonia, diarrheal illness, sepsis, cellulitis, tuberculosis, and urinary tract infections), any hospital representation, any admission, and death. The incidence of these outcomes was calculated from the cumulative number of episodes observed over 12 months or until the last day of follow-up and reported as a rate "per thousand patient-years. " Analyses of the timeto-first-event were calculated by survival analysis and compared using Kaplan-Meier and Cox regression models. Potential confounders included in the multivariable Cox models were age (<15 years, ≥15-60 years), sex, ethnicity (highland-Papuan, lowland-Papuan, non-Papuan), and presence and species of parasitemia at the time of initial admission. To control for changes in background malaria endemicity, multivariable Cox models were stratified by year of splenectomy or year of first hospital admission for controls. In the sensitivity analysis, comparisons were repeated between splenectomy patients and nonsplenectomy trauma controls.

Prospective Cohort
In a complementary prospective study, 11 patients undergoing splenectomy at an adjacent hospital (Rumah Sakit Umum Daerah [RSUD]) between July 2015 and November 2016 were followed from their day of splenectomy for 14 months. Malaria encounters were recorded by monitoring visits to healthcare facilities and conducting monthly interviews. Patients were checked for peripheral parasitemia on the day of surgery and were treated according to local treatment guidelines. Data from these patients are presented using descriptive statistics only.

Ethical Approval
This study was approved by the Ethics Committees of the University of Gadjah Mada (KE/FK/544/EC, January 2015), Indonesia, and Menzies School of Health Research, Australia (10.1397, September 2014). In the retrospective cohort, hospital records were anonymized with informed consent not requested because data were collected as part of routine hospital surveillance. Informed consent was obtained in the prospective cohort.

RESULTS
Between April 2004 and December 2013, 162 966 patients attended RSMM hospital of whom 56 458 (34.6%) were admitted at least once and 67 patients underwent splenectomy ( Figure 1). A total of 10 707 non-pregnant patients aged 12-60 years admitted on their first encounter to hospital did not have a splenectomy, of whom 1631 (15.2%) presented with a trauma-related diagnosis ( Figure 1).

Baseline Characteristics
Of the 67 splenectomized patients, 43 (64.2%) were male, 44 (65.7%) were highland-Papuans, and 4 (6.0%) were lowland-Papuans; The nonsplenectomized patients had similar demographics to that of the cases, with a median age at first admission of 26.8 (range 12-60; Table 1). The number of nonsplenectomy controls per year peaked in 2005 and declined thereafter. Thirteen splenectomy patients (19.4%) had peripheral parasitemia at the  Table 1).

Risk of Malaria in Patients With Trauma
Of the 10 707 nonsplenectomy controls, 1631 (15.2%) patients were admitted due to trauma and were considered nonsplenectomy trauma controls (Figure 1). Baseline characteristics were similar between splenectomy and trauma patients (Table 1). When compared to nonsplenectomy trauma controls, the risk of malaria after splenectomy was even greater (AHR = 6.1 [95% CI: 3.9-9.3], P < .001; Supplementary Table 2).

Malaria in a Cohort of 11 Patients Followed Prospectively From Splenectomy
Eleven patients underwent splenectomy at RSUD Hospital, all due to trauma, of which 9 (81.8%) were male, 6 (54.5%) were highland-Papuans, and 2 (18.2%) were lowland-Papuans (Supplementary Table 3). The median age at the time of splenectomy was 30 years (range . At the time of surgery, 4 had asymptomatic peripheral parasitemia by microscopy (1 P. vivax and 3 P. falciparum) and were given antimalarial treatment. In the 14 months following splenectomy, 8 of 11 (72.7%) patients returned with 18 episodes of symptomatic microscopically confirmed malaria: 8 (72.7%) patients with 12 episodes of P. vivax and 6 (54.5%) with 6 episodes of P. falciparum infection, none severe. Four of these 8 patients (50%) returned within 3 months of surgery, on day 8 (P. vivax), day 18 (P. vivax; having had asymptomatic vivax parasitemia on day 0), day 32 (P. falciparum; having had asymptomatic falciparum parasitemia on day 0) and month 3 (P. vivax), respectively. At 14 months of follow-up no patient had died.

DISCUSSION
Our large observational study quantifies the risk of morbidity and mortality in splenectomized individuals living in a malaria-endemic region. Splenectomized patients were at 5-fold higher risk of malaria within 12 months of surgery, with the risk being far greater for P. vivax compared to P. falciparum. These retrospective findings were supported by data from a prospective cohort of 11 patients followed after splenectomy. Splenectomy patients were at increased risk of representation due to any cause, admission with diarrheal illness or pneumonia, but not mortality or other causes of infection. Several case reports and small series highlight an increased risk of malaria after splenectomy; however, these reports are limited to no more than 33 splenectomized individuals [1,2,[12][13][14][22][23][24][25][26]. The current analysis was able to expand the risk analysis from an endemic area where a high number of individuals undergo splenectomy, mostly following trauma or road traffic injuries. Our study was also able to quantify, for the first time to our knowledge, the risk of vivax malaria after splenectomy and surprisingly noted that the risk was even greater than for falciparum malaria. Although our primary retrospective analysis was censored at 12 months, extending the period of follow-up to 4 years did not change the results meaningfully (Supplementary Table 4).
The greatest risk of malaria occurred in the first 3 months following splenectomy. In 2005 the incidence of malaria in Papua was approximately 876 per 1000 patient-years [18]. This comparatively low baseline incidence suggests that the rate of reinfection would have been relatively low and that episodes of malaria post-splenectomy were likely to be originating from low-level preexisting bloodstream infections. This is supported by 2 of the 4 patients with asymptomatic parasitemia in the prospective study developing clinical malaria within 32 days. Removal of the spleen impairs acquired immunity [1,7,[9][10][11], and this may induce splenectomized patients with chronic, asymptomatic parasitemia to develop higher levels of parasitemia and become symptomatic [24,25]. Alternatively, the splenectomy may result in displacement into, and/or multiplication within peripheral blood, of viable parasites that would otherwise preferentially accumulate as a hidden viable biomass in the spleen [27]. If P. vivax has a greater propensity for splenic pooling [27][28][29], this could then explain the higher risk of very early P. vivax recurrence compared to P. falciparum. Interestingly, after splenectomy a higher proportion of patients in the retrospective study had more than one episode of P. falciparum compared to P. vivax, suggesting that P. vivax attacks may have originated from preexisting infections or relapses that were then cleared, whereas P. falciparum recurrences were attributable to an increased susceptibility to new infections.
A third explanation for the increased risk of malaria is that attributable to trauma itself. The process of splenectomy is known to be followed by malaria in previously asymptomatic individuals [23][24][25][26]. Furthermore, there is an increased risk of falciparum malaria in trauma patients, a phenomenon known as postinjury malaria [30][31][32]. To address the confounding effect of trauma itself, a sensitivity analysis comparing splenectomized patients with control patients admitted due to trauma was performed. The risk of malaria was even higher in splenectomized patients, suggesting that trauma itself, at least in this population, is not a risk factor. We were unable to explore the effect of general surgery; however, earlier studies have shown that frequency of P. falciparum attacks or P. vivax relapses after surgery do not differ from the overall population or from pre-surgery [33][34][35]. Only one study reported a higher risk of P. falciparum in trauma patients undergoing primary surgery [31]. Only 11% of trauma patients are predicted to undergo surgery [36]; thus, surgery as a risk factor for malaria remains to be determined.
Previous studies have identified higher risk of sepsis including pneumonia following splenectomy [3,4,37,38]. Although pneumonia was diagnosed in 141 patients, sepsis was recorded infrequently, a likely reflection of the lack of hospital microbiological facilities. Although splenectomized patients were at more than 2.5-fold greater risk of these outcomes, this did not reach statistical significance (P = .085 and .350, respectively). Our study was also underpowered to determine an increased risk of mortality following splenectomy, with only 4 (6.0%) patients known to have died within 12 months of splenectomy. The only outcome that reached statistical significance was a 3.5-fold increased risk of diarrheal admission. Salmonellosis is caused by encapsulated bacteria, and splenectomy patients are known to be at a high risk of disease [4,39,40]; however, without an underlying microbiological diagnosis we were unable to explore this further. Our study has several limitations. Although our analysis was observational and free from coercive biases associated with many clinical study designs, our data are subject to the effects of residual confounding. Retrospective analyses were confined to RSMM presentations and did not consider presentations at other health facilities or emigration from the catchment area during the follow-up period. However, it is likely that this attrition bias would apply equally to the cases and controls. Furthermore, our findings were confirmed in the 11 patients followed prospectively. In the retrospective analysis, malaria occurring within 14 days of splenectomy was not captured and diagnosis by the attending physician was determined without microbiological testing.
In conclusion, splenectomized individuals in a malariaendemic area are at 5-fold greater risk of malaria within 12 months of surgery, with 70% of events occurring in the first 3 months. The risk is greater for P. vivax than for P. falciparum. Current guidelines recommend that patients undergoing splenectomy should be offered vaccinations and lifelong prophylactic and standby antibiotics. Our study suggests that in malaria-endemic regions these individuals should also be offered early radical cure of malaria with an artemisinin combination therapy plus, in vivax-endemic regions, 14 days primaquine. In addition, malaria prophylaxis should be offered thereafter to prevent reinfection.

Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.