Deaths From Plasmodium knowlesi Malaria: Case Series and Systematic Review

Abstract Background Plasmodium knowlesi causes severe and fatal malaria, and incidence in Southeast Asia is increasing. Factors associated with death are not clearly defined. Methods All malaria deaths in Sabah, Malaysia, from 2015 to 2017 were identified from mandatory reporting to the Sabah Department of Health. Case notes were reviewed, and a systematic review of these and all previously reported fatal P. knowlesi cases was conducted. Case fatality rates (CFRs) during 2010–2017 were calculated using incidence data from the Sabah Department of Health. Results Six malaria deaths occurred in Sabah during 2015–2017, all from P. knowlesi. Median age was 40 (range, 23–58) years; 4 cases (67%) were male. Three (50%) had significant cardiovascular comorbidities and 1 was pregnant. Delays in administering appropriate therapy contributed to 3 (50%) deaths. An additional 26 fatal cases were included in the systematic review. Among all 32 cases, 18 (56%) were male; median age was 56 (range, 23–84) years. Cardiovascular-metabolic disease, microscopic misdiagnosis, and delay in commencing intravenous treatment were identified in 11 of 32 (34%), 26 of 29 (90%), and 11 of 31 (36%) cases, respectively. The overall CFR during 2010–2017 was 2.5/1000: 6.0/1000 for women and 1.7/1000 for men (P = .01). Independent risk factors for death included female sex (odds ratio, 2.6; P = .04), and age ≥45 years (odds ratio, 4.7; P < .01). Conclusions Earlier presentation, more rapid diagnosis, and administration of intravenous artesunate may avoid fatal outcomes, particularly in females, older adults, and patients with cardiovascular comorbidities.


Systematic Review of the Literature
We searched Medline, PubMed, and Science Direct for articles published up to August 2018, containing the words "knowlesi AND death, " "knowlesi AND fatality, " or "knowlesi AND outcome, " including synonyms and Medical Subject Heading terms. Relevant references cited within these articles were also reviewed.

Case Fatality Rates
Plasmodium knowlesi CFRs in Sabah were calculated from 2010 to 2017 using malaria incidence and fatality data from the Sabah Department of Health [4,18]. Male and female CFRs were compared using Fisher exact test. Logistic regression was used to evaluate age and sex as independent risk factors for death.

Case 1 (G1)
A 32-year-old pregnant woman at 35 weeks' gestation (G6P5) presented with reduced fetal movement, 14 days of fever, and 2 days of intermittent abdominal pain and dyspnea. She was hypotensive, tachycardic, hypoxic, and tachypneic. She had hyponatremia, thrombocytopenia, and acute kidney injury (AKI). She was intubated, ventilated, and given IV ceftriaxone, sodium bicarbonate, and ionotropic support. A chest radiograph showed diffuse interstitial infiltrates consistent with acute respiratory distress syndrome (ARDS). A blood film taken 4 hours after admission was reported as P. knowlesi, with 22 400 parasites/µL. Oral artemether-lumefantrine and doxycycline were given, and tertiary hospital transfer was arranged for IV artesunate; however, the patient had a cardiac arrest while awaiting transfer. Admission blood cultures were negative.

Case 2 (G2)
A 50-year-old male Indonesian palm-oil plantation worker presented to a district hospital with 1 week of fever, rigors, headache, epigastric pain, and diarrhea. Physical examination revealed hepatomegaly (3 cm), but was otherwise unremarkable. He had severe AKI, hyponatremia, and thrombocytopenia. A blood film was reported as P. knowlesi with 71 939 parasites/ µL. Chest radiography showed generalized interstitial infiltrates. He was diagnosed with severe malaria and commenced on IV artesunate, and a tertiary hospital transfer was arranged; however, he arrested en route. This case has been previously documented [10].

Case 3 (G3)
A 37-year-old woman with a history of gestational hypertension presented to a district hospital with 4 days of fever, rigors, epigastric pain, dizziness, and vomiting. She was hypotensive, tachycardic, and had epigastric tenderness. The initial diagnosis was dyspepsia, and she received ranitidine, prochlorperazine, and fluid resuscitation. AKI was noted 2 hours after admission, and IV ceftriaxone was initiated for presumed sepsis. Twelve hours after hospitalization, she became tachypneic, with reduced consciousness (Glasgow Coma Scale reported as 9/15), and severe metabolic acidosis (arterial pH, 7.02; bicarbonate, 2.5 mEq/L). She underwent intubation, ventilation, and hemodialysis and received IV sodium bicarbonate and imipenem. The patient deteriorated further on day 2, with transaminitis (alanine aminotransferase [ALT] level, 426 U/L; aspartate aminotransferase [AST] level, 1481 U/L), metabolic acidosis, and coagulopathy (international normalized ratio [INR], 1.9; prothrombin time, 52 seconds) requiring fresh frozen plasma. The patient arrested 43 hours after admission. Blood films taken prior to death were reported as P. malariae with 2285 parasites/ µL. It was later noted that an admission blood film had been reported as "P. falciparum 3+. " No antimalarial treatment had been given. Blood cultures were negative.

Case 4 (G4)
A 32-year-old Filipino man presented with 5 days of fever, cough, and coryzal symptoms. He was tachycardic, but physical examination was otherwise unremarkable. He was given fluids and acetaminophen (paracetamol). A blood smear was reported as "P. malariae 1+" with AKI, hyponatremia, and thrombocytopenia also noted. Oral artemether-lumefantrine was given. Twenty-two hours after admission, the patient had hemoptysis with tachypnea and hypoxia. A chest radiograph showed generalized interstitial infiltrates. A diagnosis of severe P. malariae with ARDS was made and IV artesunate was initiated. After further hemoptysis, a bedside ultrasound showed pleural and pericardial effusions and noninvasive ventilation was initiated. At 48 hours, massive hemoptysis occurred. He required intensive care unit admission, intubation, ventilation, inotropic support, hemodialysis for metabolic acidosis and uremia, and red cell and fresh frozen plasma transfusions. Echocardiography (ECG) showed severe mitral stenosis. Admission blood cultures were negative. He died on day 4, with cause of death recorded as severe malaria with pulmonary hemorrhage and underlying mitral stenosis.

Case 5 (G5)
A 42-year-old fisherman with morbid obesity, obstructive sleep apnea, and congestive cardiac failure was referred to a district hospital from a peripheral clinic with 9 days of fever, and 3 days of dyspnea, epigastric pain, and vomiting. The clinic blood film was reported as "complicated malaria with hyperparasitemia"; Plasmodium species was not specified. On admission he was tachycardic, hypotensive, and hypoxic. ECG showed atrial fibrillation at 160 beats per minute. He received IV artesunate,  One fatal case of microscopy-diagnosed "Plasmodium malariae" was described in this series, but has not been included in the review due to lack of polymerase chain reaction confirmation. f This patient was also described in a separate case report [13]. g Thirty-five weeks' gestation at presentation. h IV therapy not available at presenting hospital. i This patient was previously reported in a prospective observational study [10]. j Values are presented as the median.
fluid resuscitation, and high-flow oxygen. He was hyponatremic, with AKI, hyperglycemia, and a compensated metabolic acidosis. A blood film was reported as "P. malariae 4+. " He required intubation, hemodialysis, IV insulin, digoxin, and inotropic support. Seven hours postadmission, ECG showed cardiac ischemia, and at 15 hours the patient arrested. Admission blood cultures were negative. Cause of death was recorded as (1) malaria with hyperparasitemia, and (2) decompensated cardiac failure, with likely underlying cardiomyopathy.

Case 6 (G6)
A 58-year-old Caucasian expatriate man with a history of ischemic heart disease, left bundle-branch block, hypertension, and paroxysmal atrial fibrillation presented to a private hospital with 9 days of fever, palpitations, and lethargy. He was hypotensive, hypoxic, and tachypneic. A blood smear was reported as "heavy infection, likely P. falciparum. " He was anemic, hyperbilirubinemic, and hyponatremic, with AKI and metabolic acidosis. He was commenced on ionotropic support and transferred to a tertiary public hospital for IV artesunate. There he also received ceftriaxone; hydrocortisone; insulin infusion for hyperglycemia; cardioversion for ventricular tachycardia; and, later, intubation, ventilation, and continuous venovenous hemofiltration. A blood film on day 1 was reported as P. knowlesi with 246 100 parasites/µL. On day 2, he had worsening metabolic acidosis (lactate, 11.7 mmol/L; pH, 7.13; bicarbonate, 10.4 mEq/L), coagulopathy (INR, 4.53), and transaminitis (AST, >4200 U/L; ALT, 1820 U/L), with a bilirubin of 247 μmol/L, and was anuric. He died on day 3, with cause of death recorded as severe knowlesi malaria with multiorgan failure.

Systematic Review of P. knowlesi Fatalities
Ten original research articles reporting P. knowlesi fatalities were identified (Figure 1). Thirty PCR-confirmed P. knowlesi deaths were reported: 19 from Sabah and 11 from Sarawak, Malaysia, all occurring during 2001-2014. Four of the Sarawak deaths were excluded from further review due to insufficient individual patient data [15]. Demographic, clinical, and laboratory details for each case are summarized in Tables 1-3, with severity criteria shown in Table 4. Combining these 26 previously reported cases with the 2015-2017 series, median age was 56 (range, 23-84) years, and 18 (56%) were male. Female fatal cases were older than male fatal cases (median age, 62 vs 51 years, respectively), although this was not statistically significant. Median duration of fever was 6 (interquartile range, 4-7) days, with no difference between women and men. At presentation, 30 of 32 (94%) patients met the World Health Organization (WHO) criteria for severe malaria, although only 19 of these 30 (63%) were recognized as having severe malaria. Of those meeting severity criteria, the median number of criteria was 3 (range, 1-7), with the most frequent criteria being respiratory distress (n = 19/32 [59%]), jaundice (n = 18/25 [72%]), and severe AKI (n = 19/27 [70%]). Abdominal pain was reported in 17 of 26 (65%) cases. Decreased conscious state was reported on presentation in 1 case; however, alternative causes were not excluded [4]. Cardiovascular-metabolic comorbidities were a notable factor of previously reported cases as well as in the current series. Overall, hypertension was reported in 8 of 32 (25%) cases and stress hyperglycemia or diabetes mellitus in 12%. Among all fatal cases, species diagnosis on admission microscopy was incorrect in 26 of 29 (90%), including 20 (69%) diagnosed as P. malariae, 4 (14%) as P. falciparum, and 2 (7%) as Plasmodium vivax. Thrombocytopenia was universal on presentation, as was elevated creatinine (Table 3). All patients with an available serum sodium result were hyponatremic. No patient had severe anemia at presentation. Intravenous antimalarial treatment was administered on presentation in 17 of 21 (81%) cases with severe malaria on presentation. Median time to death was 41 (range, 1-316) hours.
Abbreviations: AKI, acute kidney injury; Hyper, hyperglycemia; N, no; NR, not reported; Y, yes. a Not reported to date in knowlesi malaria. b Alternative causes for reduced GCS not excluded.

DISCUSSION
We describe 6 malaria deaths in Sabah during 2015-2017, and review 26 previously reported P. knowlesi deaths. Plasmodium knowlesi accounted for all recent malaria deaths in Sabah, in contrast to previous reports during 2010-2014 where 13 of 29 (44%) deaths were attributed to P. knowlesi [4,12]. This increase in proportion of fatal cases attributed to P. knowlesi is consistent with the decline of falciparum and vivax malaria in Sabah, and the ongoing increase in cases of P. knowlesi [4,18]. The fact that the total number of P. knowlesi deaths per year has remained relatively stable despite this increasing incidence is consistent with previously reported declining CFRs in Sabah [4], likely reflecting ongoing increases in awareness and improved management of severe knowlesi malaria. It is notable that all reported fatal cases were adults. This is consistent with the absence of severe knowlesi malaria in children [10,19], and is in marked contrast to the predominance of severe and fatal malaria from P. falciparum and P. vivax in the pediatric age group [20]. Older age is a known risk factor for severe knowlesi malaria [8,10,17], and in this study age ≥45 years was associated with a 5-fold increase in risk of death. It was also notable that nearly half of all reported fatal cases were women. This contrasts with the predominance of men in studies of nonfatal knowlesi malaria, with men consistently accounting for 75%-80% of cases [21,22].  [4,9,12,14], 2 prospective observational studies [10,11], 1 case-control study [15], 2 case reports [13,16], and 1 review of autopsies [35]. b Four reported deaths were excluded from further review as sufficient individual patient data were not available [15]. We previously hypothesized that the higher CFR in women with knowlesi malaria was due in part to their older median age [4]. However, here we now demonstrate that women are more than twice as likely to die from knowlesi malaria even after adjusting for age. There was no difference in parasitemias between male and female fatal cases, nor any difference in fever duration, suggesting that the higher mortality was not due to delays in health-seeking behavior. A higher risk of death in females has also been noted in other severe infections [23][24][25]. Further studies are required to investigate the pathogenic or other factors that may account for this increased risk [24]. Respiratory distress and AKI were the most common severity criteria on presentation in fatal cases. Abdominal pain, previously reported as a risk factor for severe knowlesi malaria [10], was present in 53% of all cases. Potential explanations include gut ischemia from microvascular accumulation of parasitized red cells and/or gastric ulceration [14]. Thrombocytopenia was universal among fatal cases. This may relate to the higher proportion of platelet binding to infected red cells found in P. knowlesi compared with other species [26]. Hyponatremia, a known metabolic abnormality in severe falciparum and knowlesi malaria [8,10,27], was present in all fatal cases. Hypoglycemia occurred in 15% of previously reported fatal cases; however, in this series, 2 cases of hyperglycemia associated with death also occurred. This may relate to the older age of knowlesi-infected patients with severe malaria [8,17], and rising prevalence of metabolic syndrome with age.
Comorbidities are a notable feature in fatal knowlesi malaria. Half of the newly described cases had underlying cardiovascular-metabolic disease (including severe mitral stenosis, heart failure, and morbid obesity), and hypertension and hyperglycemia or diabetes mellitus were also common in the previously reported cases. Cardiovascular disease and metabolic syndrome are known to be associated with increased systemic inflammation, endothelial activation, and microvascular dysfunction [28], all features of severe knowlesi malaria [17]. It is thus plausible that cardiovascular-metabolic disease would increase the risk of severe knowlesi malaria, as has been reported in falciparum malaria [29].
This series includes the first report of fatal knowlesi malaria in pregnancy, occurring at 35 weeks' gestation. Plasmodium knowlesi is relatively rare in pregnancy, with only 5 cases reported previously [9,30]. Pregnancy is known to increase risk of severe maternal disease in falciparum malaria [20]. Of the 5 previously described cases of P. knowlesi in pregnancy, 1 had severe malaria with intrauterine death, 2 had moderate anemia, and 1 delivered a preterm low-birth-weight infant [9,30]. With the current fatal case, 2 of 6 (33%) reported cases of P. knowlesi in pregnancy have resulted in severe maternal malaria and fetal death. Prospective studies are required to further evaluate the true risks and consequences of knowlesi malaria in pregnancy. Several health-system issues contributed to deaths in this newly reported series. In 3 cases, the diagnosis of severe malaria was delayed despite the presence of compatible clinical features, and in another 2, IV artesunate was not available. Half of the cases were non-Malaysian citizens, which, together with the long duration of fever in this series (median, 8 days), suggests possible barriers to accessing healthcare. These findings highlight the need to evaluate strategies to improve healthcare access in endemic areas, in addition to ensuring availability of appropriate antimalarials at all facilities. In the combined analysis, 90% of fatal cases had an alternative Plasmodium species diagnosed on admission microscopy. This is consistent with the inability of routine microscopy to reliably distinguish P. knowlesi from other species [31], which, combined with the poor sensitivity and specificity of available rapid diagnostic tests [32,33], highlights the importance of a unified treatment strategy of IV artesunate for severe malaria from any species [20,34].
A limitation of this study was the retrospective nature of the case series, resulting in either incomplete or unvalidated data. In particular, parasite counts were surprisingly low in some cases (G4 and G3), and it is possible that blood films may have been misreported. Second, the Department of Health data used to calculate CFRs did not include parasite counts, and hence the contribution of parasitemia to age and sex as risk factors for death could not be evaluated. Plasmodium knowlesi parasitemia is known to increase with age; however, age has been shown to be an independent risk factor for severe knowlesi malaria [10,17], and hence would be expected to be an independent risk factor for death. In the largest prospective studies conducted to date [10,17], an independent association between parasitemia and sex was not found; thus, a difference in parasitemia is unlikely to account for the increased CFR in women. Finally, it is possible that not all cases of P. knowlesi are notified, so the CFRs reported in this manuscript may overestimate the true CFRs.
In conclusion, this review highlights the potential for poor outcomes from knowlesi malaria, and identifies female sex, age ≥45 years, and comorbidities as important associated factors. With the near-elimination of falciparum and vivax malaria in Sabah, Malaysia, health systems must maintain efforts to promptly diagnose and treat patients with knowlesi malaria to avoid fatal outcomes, particularly in these at-risk groups. This will require ongoing surveillance and efforts to increase awareness among communities and health providers, as well as improving access to healthcare and ensuring availability of appropriate antimalarials at all health facilities.