Mosquito vectors of lymphatic and zoonotic filariasis in Malaysia: a systematic review and proportionate meta-analysis of infection rates

Abstract

Lymphatic and zoonotic filariasis in Malaysia are transmitted by multiple mosquito species, but the infection rates in these vectors have not been quantitatively reviewed. This systematic review and meta-analysis of proportions compiled species-level infection data for Malaysian mosquitoes infected with Brugia malayi, Brugia pahangi, Wuchereria bancrofti and Dirofilaria immitis. Searches of PubMed, Scopus and Web of Science identified 297 records; 12 studies met the predefined inclusion criteria, yielding 42 mosquito–parasite records involving Mansonia, Anopheles, Armigeres, Aedes and Culex. Infection rates were pooled separately for natural and experimental infections using generalised linear mixed-effects models with logit transformation. For B. malayi, natural infection rates in Mansonia and Anopheles were very low (pooled 0.35%; 0.02–4.46% across species) despite large sample size, whereas experimental infections were far higher (pooled 72.4%). Wuchereria bancrofti showed low natural infection rates (pooled 2.1%; 0.9–6.7% across species) and higher experimental infection (pooled 36.8%). In contrast, B. pahangi showed high natural infection in Armigeres subalbatus (pooled 7.2%; range 3.9–14.4%), with experimental infection of 45.0%. Dirofilaria immitis showed very low natural infection (pooled 0.06%) but comparatively high experimental infection (pooled 29.1%), reaching 55–69% in Aedes togoi. Natural and experimental infection rates differed significantly for every parasite (tests for subgroup differences, p < 0.0001). Overall, classical human filariasis parasites now circulate at low levels in their traditional Mansonia and Anopheles vectors, while zoonotic parasites are maintained in peri-domestic mosquito–animal cycles. These findings highlight the need for integrated One Health strategies combining mass drug administration, peri-domestic vector control, improved diagnostics and veterinary interventions to achieve sustainable filariasis elimination in Malaysia.

Supplementary Information

The online version contains supplementary material available at 10.1186/s41182-026-01055-w.

Introduction

Lymphatic filariasis (LF), commonly known as elephantiasis, is a neglected tropical disease caused by infection with thread-like nematode parasites from the family Filarioidea, primarily Wuchereria bancrofti, Brugia malayi and Brugia timori [1]. The disease is transmitted to humans through the bite of infected mosquitoes, which carry filarial parasites, and can lead to extreme swelling of the arms, legs, and other body parts [2]. According to the Global Burden of Disease 2024 report by the Institute for Health Metrics and Evaluation (IHME), the global prevalence of lymphatic filariasis (LF) in 2021 was estimated at 56.90 million individuals (Accessed November 15th, 2025) [3]. In Malaysia, approximately 338,593 cumulative number individuals were identified as positive for LF as of 2021. Furthermore, for 2025, national surveillance data are drawn from June, July and August–September editions of the Epidemiology Bulletin of Disease Control Division, Malaysia. In these bulletins the number of reported LF cases showed notable monthly variation, with 43 cases recorded in June, a marked increase to 122 cases in July, followed by 105 cases in August and a subsequent decline to 35 cases in September [4–6].

For the past decades, Malaysia has made substantial progress towards LF elimination through mass drug administration (MDA) and integrated control strategies (i.e. vector control and strengthened surveillance); however, the persistence of transmission in certain endemic foci indicates that elimination has not yet been achieved. According to the Annual report 2023 from the National Lymphatic Filariasis Elimination Programme, Ministry of Health Malaysia [7], LF prevalence has been successfully reduced to below 2% across all 81 (100%) endemic sub-districts in Peninsular Malaysia. Nevertheless, 5 of the 46 (11%) endemic sub-districts in Sabah and Sarawak remain above the target threshold. To accelerate elimination in these remaining foci, active surveillance and MDA with a triple-drug regimen comprising ivermectin, diethylcarbamazine (DEC) and albendazole are being implemented, with the goal of achieving less than 2% prevalence nationwide. [7]

Beyond human lymphatic filariasis, Malaysia also sustains zoonotic filarial transmission of growing public-health concern. Brugia pahangi, a filarial nematode of domestic cats and dogs, and Dirofilaria immitis, the canine heartworm, circulate in peri-domestic mosquito populations, and sporadic zoonotic B. pahangi infections have been documented in humans in Peninsular Malaysia [8–10]. Because these animal-derived parasites share mosquito vectors and larval habitats with human Brugia, their transmission is epidemiologically intertwined with lymphatic filariasis elimination, and mosquito infection with zoonotic filariae can complicate molecular xenomonitoring and diagnosis. Accounting for both anthroponotic and zoonotic cycles is, therefore, essential when characterising the mosquito vectors responsible for filarial transmission in Malaysia.

The low-level transmission that persists in the remaining foci of Sabah and Sarawak is best understood as residual transmission continuing after decades of sustained control, rather than genuine re-emergence of previously eliminated infection; the principal re-emergence concern instead relates to the zoonotic B. pahangi and D. immitis cycles described above, which are not targeted by human mass drug administration [11]. Quantifying which mosquito species carry these parasites, and at what infection rates, is a prerequisite for effective surveillance and targeted vector control. Although vector distribution, behaviour and competence also shape transmission, the quantitative evidence available across the Malaysian literature is largely confined to mosquito infection rates; this review, therefore, focuses on synthesising these infection rates, which have not previously been compiled or pooled across the published studies.

Previous studies in Malaysia have described individual vectors, including swamp breeding Mansonia spp. and multiple Anopheles species for B. malayi and W. bancrofti [12–14] and more recently Armigeres subalbatus, Culex quinquefasciatus and Aedes togoi for zoonotic B. pahangi [8–10]. However, these findings have not been synthesised using systematic methods or quantitative pooling of infection rate. This study aims to: (i) systematically identify and critically appraise all studies reporting mosquito infection rates for lymphatic and zoonotic filarial parasites in Malaysia; (ii) quantitatively pool species-level natural and experimental infection rates for the principal vector species using a meta-analysis of proportions; and (iii) map how these infection patterns relate to the diagnostic methods and vector-control measures documented in the Malaysian literature, thereby situating the pooled estimates within the national elimination context.

Methods

Systematic review and proportionate meta-analysis

This review assessed published research on filarial infections in Malaysian mosquito vectors associated with both human and zoonotic filariasis through a systematic synthesis and meta-analysis of proportions, adhering to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. The primary focus was on the infection rates of B. malayi, B. pahangi, W. bancrofti and D. immitis in identified mosquito species. The review also encapsulated the diagnostic techniques employed for identifying filarial infections in vectors, as well as an overview of both historical and contemporary vector-control strategies, the operational challenges faced in vector management and prospective approaches for surveillance and control in Malaysia. A registration with the International Prospective Register of Systematic Reviews (PROSPERO) was submitted (Submission ID: 1269415); eligibility criteria, search strategy, extraction and analysis plans were established beforehand and implemented consistently.

Search strategy

A systematic search was conducted in PubMed, Web of Science and Scopus. The core search string combined terms for filariasis, mosquitoes and Malaysia, for example: “(lymphatic filariasis OR filariasis OR filaria OR filarial OR Brugia OR Brugia malayi OR Brugia pahangi OR Wuchereria bancrofti OR Dirofilaria immitis) and (mosquito* OR vector* OR Mansonia OR Anopheles OR Aedes OR Culex OR Armigeres) and (Malaysia OR Peninsular Malaysia OR Sabah OR Sarawak OR Borneo)”. Searches were limited to studies conducted in Malaysia, published in English between 1940 and 2025. Although the bibliographic databases (PubMed, Scopus, Web of Science) provide largely overlapping coverage of journal articles, publisher platforms also expose conference proceedings, supplements and grey-literature items not consistently captured elsewhere. Duplicate records arising from the substantial overlap between sources were managed in a two-step process: (i) automated de-duplication in EndNote 21 using DOI, then title plus first author plus publication year and (ii) manual confirmation of residual potential duplicates during title and abstract screening, with the earliest or most complete version retained. After de-duplication, 171 unique records remained from 297 retrieved. Titles and abstracts were screened, and accessible full texts meeting inclusion criteria were retrieved and assessed for eligibility. The final database search was conducted on 15th December 2025 and full electronic search strategies for each database, including all search terms, limits and filters, are provided in Supplementary Table S1.

Eligibility criteria

Studies were eligible if they met all of the following criteria:

• Conducted within Malaysia.

• Mosquitoes identified to the species level.

• Filarial parasites identified as B. malayi, B. pahangi, W. bancrofti or D. immitis.

• Reported both the number of mosquitoes examined and the number infected for each species–parasite combination.

• Experimental infection studies were included only when the mosquito populations originated from Malaysia or were established from Malaysian field collections.

Both natural infection studies (wild-caught mosquitoes) and experimental infection studies (laboratory exposure to infected hosts or parasites) were eligible. Studies that reported only infection percentages without denominators, pooled multiple mosquito species, or did not specify the filarial species were excluded.

Data extraction

All eligible studies were entered into a standardised spreadsheet. For each mosquito–parasite combination the following variables were extracted: study identifier, authors and year, publication year, state or region, ecosystem type, mosquito genus and species, filarial species, detection method (for example, dissection or PCR), study design (natural or experimental infection), number examined, number infected, larval stage detected and short notes on sampling or experimental conditions. Data extraction was performed independently by two reviewers using a standardised form and discrepancies were resolved by checking the original articles and reaching consensus.

Study design was recorded as “NI” for natural infections and “EI” for experimental infections. For plotting, a combined label was created in the format “NI- Mansonia uniformis-Hodgkin, 1941” or “EI- Aedes togoi-Vythilingam et al., 2005b”. Data were checked independently by two readers and discrepancies were resolved by re-examining the original publications. The primary outcome for this review was mosquito infection rate, defined as the proportion of examined mosquitoes of a given species that were infected with any filarial stage or tested positive by molecular methods. No additional outcome measures (e.g. intensity of infection) were extracted.

Risk-of-bias assessment

Risk of bias for the included mosquito infection studies was assessed at the study level using an adapted Joanna Briggs Institute (JBI) critical appraisal checklist for prevalence studies. Eight domains were evaluated: (1) clearly defined inclusion criteria; (2) detailed description of study setting and mosquito collections; (3) appropriate and clearly described sampling methods; (4) adequate or justified sample size; (5) valid and reliable identification of mosquito and filarial species; (6) standard, objective criteria for defining infection (e.g. specified larval stages or validated PCR); (7) appropriate statistical analysis of infection rates; and (8) adequate reporting of denominators and outcomes (numbers examined and infected). Each item was rated “Yes”, “No” or “Unclear”and an overall judgement of low, moderate or high risk of bias was assigned for each study based on the pattern of domain scores. Two reviewers performed the assessments independently and resolved disagreements by discussion. Brief justifications for each overall judgement are provided in Supplementary Table S2. Risk-of-bias (ROB) categories (L = low; M = moderate; H = high) were also indicated next to study labels in forest plots to aid interpretation. Detailed item-level risk-of-bias ratings for each study, based on the adapted JBI prevalence checklist, are provided in Supplementary Table S3.

Statistical analysis

Infection rate was defined as the proportion of examined mosquitoes of a given species that were infected with any larval stage (L1–L3) or tested positive by molecular methods. Meta-analyses of single proportions were conducted using the metaprop function in R’s meta package. To stabilise variances, a logit transformation was used and generalised linear mixed‑effects models were fitted with maximum‑likelihood estimation of the between‑study variance (sm = “PLOGIT”, method = “GLMM”, method.tau = “ML”). A continuity correction of 0.5 was applied when needed for zero events.

Separate meta-analyses were carried out for B. malayi, B. pahangi, W. bancrofti and D. immitis. All available mosquito species for a given parasite were included in the primary model. Because data for some vector–parasite combinations were sparse, subgroup analyses by mosquito species were treated as descriptive. Heterogeneity was summarised with τ² and I². Forest plots present study-specific proportions with 95% confidence intervals, together with pooled common-effect and random-effects estimates.

No formal subgroup analyses, meta-regression or sensitivity analyses were conducted, because the number of studies within each parasite–vector grouping was small and study designs were heterogeneous. Heterogeneity was, therefore, described using I² and τ², and its implications were interpreted qualitatively.

Reporting bias assessment

No formal assessment of reporting bias (e.g. funnel plots or statistical tests for small study effects) was undertaken, because each parasite-specific synthesis included too few studies for such methods to be informative.

Assessment of certainty of the evidence

A formal Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework was not applied, because the included studies were few, highly heterogeneous and entomological in nature rather than clinical trials. Instead, certainty in the body of evidence for each parasite–vector grouping was assessed qualitatively. The following factors were considered: (i) precision of pooled infection-rate estimates and their 95% confidence intervals; (ii) consistency of patterns across studies and mosquito species; (iii) risk of bias based on the adapted JBI checklist; and (iv) directness of the evidence in relation to current Malaysian transmission settings (e.g. natural versus experimental infection; historical versus recent data). For each parasite, two reviewers independently judged overall certainty as ‘higher’, ‘moderate’ or ‘lower’ and resolved any differences by discussion. These qualitative certainty judgments are reported narratively in the Discussion.

Results

Databases search outcome and study characteristics

The search yielded 297 records: 144 from PubMed, 102 from Scopus and 51 from Web of Science. After removing 126 duplicates, 171 unique titles and abstracts were screened. Records clearly unrelated to filarial infection in Malaysian mosquitoes were excluded, leaving 30 reports for full‑text assessment. 10 full texts could not be obtained. Of the remaining 20, eight were excluded, because they reported data from other countries or did not provide extractable infection counts.

Of the 297 records identified, 12 studies met the inclusion criteria and contributed mosquito infection data to the meta-analysis (Fig. 1). These span 1940–2025 and cover Peninsular Malaysia, Sabah and Sarawak, with a range of rural, swamp forest-edge, suburban and urban settings.

Fig. 1 .

PRISMA flow diagram of the study selection process for the systematic review of mosquito vectors of filarial parasites in Malaysia

Table 1 summarises the characteristics of the included studies. Early work from the 1940s–1980s focused on B. malayi and W. bancrofti in swamp and rural habitats, with Mansonia spp. and Anopheles spp. as principal vectors and dissection as the sole diagnostic method. Later studies from the 1990s onwards increasingly investigated zoonotic B. pahangi and D. immitis in suburban Kuala Lumpur and Selangor, involving Armigeres spp., Aedes spp. and Culex spp. species and often using PCR alongside dissection. Only mosquito records with explicit denominators (numbers examined and infected) were included in the meta-analysis.

Table 1.

Description of included mosquito infection studies with parasite species, main vectors, detection method, study location and epidemiological setting

SN	Author (year)	Filarial species reported	Principal mosquito vectors	Detection method	Study location/state (region)	Setting	Study design (NI/EI)
1	Hodgkin (1941) [15]	B. malayi	Mn. uniformis	Dissection	Peninsular Malaysia	Endemic rural	NI
2	Reid & Ganapathipillai (1962) [16]	B. malayi; Dirofilaria sp.; Setaria sp.	Mn. uniformis, Mn. indiana, Mn. dives, An. campestris, An. nigerrimus, An. peditaeniatus (and others)	Dissection	Peninsular Malaysia	Endemic suburban	NI/EI
3	Hii et al. (1985a) [12]	W. bancrofti	An. balabacensis, An. kochi, Ae. togoi, Cx. quinquefasciatus	Dissection	Sabah	Endemic swamp	NI/EI
4	Hii et al. (1985b) [17]	W. bancrofti	An. balabacensis, An. flavirostris	Dissection	Sabah	Endemic suburban	NI
5	Chiang et al. (1986) [13]	B. malayi	An. sinensis	Dissection	Peninsular Malaysia	Endemic rural	NI
6	Vythilingam et al. (1996) [14]	B. malayi	An. donaldi, An. nigerrimus, Mn. dives, Mn. uniformis	Dissection	Peninsular Malaysia	Endemic suburban	NI
7	Vythilingam et al. (2005a) [18]	W. bancrofti	Cx. quinquefasciatus	Dissection	Peninsular Malaysia	Urban endemic	EI
8	Vythilingam et al. (2005b) [19]	D. immitis	Ae. albopictus, Ae. aegypti, Ae. togoi, Ar. subalbatus, Cx. quinquefasciatus	Dissection	Peninsular Malaysia	Urban zoonotic	EI
9	Tan et al. (2011) [8]	B. pahangi	Ar. subalbatus	Dissection; PCR	Peninsular Malaysia	Suburban zoonotic	NI
10	Muslim et al. (2013a) [9]	B. pahangi	Ar. subalbatus	Dissection; PCR	Peninsular Malaysia	Suburban zoonotic	NI
11	Muslim et al. (2013b) [20]	B. pahangi	Ar. subalbatus	Dissection; PCR	Peninsular Malaysia	Suburban zoonotic	NI
12	Vinnie-Siow et al. (2022) [10]	B. pahangi; D. immitis	Ae. togoi, Cx. quinquefasciatus	Dissection; PCR	Peninsular Malaysia	Suburban zoonotic	EI

NI Natural infection, EI Experimental Infection

Evolution of mosquito vectors in the dataset

Across the 12 studies, the composition of vector species reflects a chronological shift in filariasis ecology. Historical investigations in Penang and north–west Malaya identified Mansonia uniformis, Mansonia indiana and Mansonia dives as important vectors of B. malayi, supplemented by Anopheles campestris and other Anopheles species. Subsequent work in Sabah and Sarawak identified Anopheles balabacencis, Anopheles flavirostris, Anopheles barbirostris, Anopheles donaldi and other species as vectors of W. bancrofti in rural, coastal and forest-fringe communities.

From the 2000s onwards, additional culicine genera emerged as vectors. Suburban studies incriminated Armigeres subalbatus as a natural vector of B. pahangi and showed that Ae. togoi and Culex quinquefasciatus can transmit B. pahangi and D. immitis in experimental infections. This transition from Mansonia spp. to Anopheles spp. systems in rural swamps to peri-domestic Armigeres spp., Aedes spp. and Culex spp. in suburban and urban environments is reflected in the infection-rate patterns in figures below (Figs. 2, 3, 4, 5).

Fig. 2.

Meta-analysis of proportions forest plot of mosquito infection rates with B. malayi in Malaysia

Fig. 3.

Meta-analysis of proportions forest plot of mosquito infection rates with W. bancrofti in Malaysia

Fig. 4.

Meta-analysis of proportions forest plot of mosquito infection rates with B. pahangi in Malaysia

Fig. 5.

Meta-analysis of proportions forest plot of mosquito infection rates with D. immitis in Malaysia

Pooled infection rates by parasite and vector group

Pooled analyses of mosquito–parasite records from 12 studies (1941–2022) using logit-transformed generalised linear mixed models (GLMMs), with natural and experimental infections modelled as separate subgroups, revealed substantial heterogeneity across ecological settings and study designs (Figs. 2–5). Tests for subgroup differences between natural and experimental infection were significant for the three parasites (p < 0.0001) assessed under both conditions (B. malayi, B. pahangi and W. bancrofti), whereas D. immitis was reported from experimental infection only; Tables 2 and 3 summarise the natural and experimental infection rates, respectively.

Table 2.

Natural infection rates of filarial parasites in mosquitoes in Malaysia

Mosquito species	Infected/examined	Rate (%)	95% CI (Wilson)	Source	Year/RoB
Brugia malayi
Mn. uniformis	1/658	0.15	0.03–0.86	Hodgkin (1941) [15]	1941/M
An. campestris	19/3573	0.53	0.34–0.83	Reid & Ganapathipillai (1962) [16]	1962/M
Mn. indiana	2/4424	0.05	0.01–0.16		1962/M
Mn. uniformis	1/4316	0.02	0.00–0.13		1962/M
An. sinensis	1/142	0.70	0.12–3.88	Chiang et al. (1986) [13]	1986/M
An. donaldi	23/516	4.46	2.99–6.60	Vythilingam et al. (1996) [14]	1996/L
An. nigerrimus	7/476	1.47	0.71–3.00		1996/L
Mn. dives	2/188	1.06	0.29–3.80		1996/L
Mn. uniformis	1/230	0.43	0.08–2.42		1996/L
Random-effects subtotal (I2 = 91.6%)		0.35	0.11–1.09
Brugia pahangi
Ar. subalbatus	54/801	6.74	5.20–8.69	Tan et al. (2011) [8]	2011/L
Ar. subalbatus	48/334	14.37	11.01–18.54	Muslim et al. (2013a) [9]	2013/M
Ar. subalbatus	62/1599	3.88	3.04–4.94	Muslim et al. (2013b) [20]	2013/M
Random-effects subtotal (I2 = 96.0%)		7.21	3.86–13.08
Wuchereria bancrofti
An. balabacensis	1/17	5.88	1.05–26.98	Hii et al. (1985a) [12]	1985/L
An. balabacensis	4/365	1.10	0.43–2.78	Hii et al. (1985b) [17]	1985/L
Cx. quinquefasciatus	11/164	6.71	3.79–11.61		1985/L
An. flavirostris	9/1001	0.90	0.47–1.70		1985/L
Random-effects subtotal (I2 = 87.5%)		2.07	0.76–5.51

Table 3.

Experimental infection rates of filarial parasites in mosquitoes in Malaysia

Mosquito species	Infected/examined	Rate (%)	95% CI (Wilson)	Source	Year/RoB
Brugia malayi
An. baezai	11/32	34.38	20.41–51.69	Reid & Ganapathipillai (1962) [16]	1962/M
An. barbirostris	63/72	87.50	77.92–93.28		1962/M
An. campestris	22/24	91.67	74.15–97.68		1962/M
An. crawfordi	5/5	100.00	56.55–100.00		1962/M
An. indiensis	9/31	29.03	16.10–46.59		1962/M
An. lesteri	37/93	39.78	30.43–49.95		1962/M
An. nigerrimus	76/146	52.05	44.00–60.00		1962/M
An. peditaeniatus	31/34	91.18	77.04–96.95		1962/M
An. sinensis	3/99	3.03	1.04–8.53		1962/M
Mn. annulata	1/10	10.00	1.79–40.42		1962/M
Mn. annulifera	71/76	93.42	85.51–97.16		1962/M
Mn. dives	1/1	100.00	20.65–100.00		1962/M
Mn. indiana	34/39	87.18	73.29–94.40		1962/M
Mn. uniformis	22/22	100.00	85.13–100.00		1962/M
Random-effects subtotal (I2 = 91.5%)		72.43	42.38–90.37
Brugia pahangi
Ae. togoi	8/16	50.00	28.00–72.00	Vinnie-Siow et al. (2022) [10]	2022/L
Cx. quinquefasciatus	1/4	25.00	4.56–69.94		2022/L
Random-effects subtotal (I2 = 0.0%)		45.00	25.32–66.38
Dirofilaria immitis
Ae. albopictus	1/8	12.50	2.24–47.09	Vythilingam et al. (2005b) [19]	2005/M
Ar. subalbatus	1/19	5.26	0.94–24.64		2005/M
Ae. aegypti	39/75	52.00	40.87–62.93		2005/M
Ae. togoi	21/38	55.26	39.71–69.85		2005/M
Cx. quinquefasciatus	2/21	9.52	2.65–28.91		2005/M
Ae. togoi	9/13	69.23	42.37–87.32	Vinnie-Siow et al. (2022) [10]	2022/L
Random-effects subtotal (I2 = 77.8%)		29.05	11.20–57.06
Wuchereria bancrofti
An. kochi	1/5	20.00	3.62–62.45	Hii et al. (1985a) [12]	1985/L
An. balabacensis	20/48	41.67	28.85–55.72		1985/L
Ae. togoi	13/38	34.21	21.21–50.11		1985/L
Cx. quinquefasciatus	5/15	33.33	15.18–58.29	Vythilingam et al. (2005a) [18]	2005/L
Random-effects subtotal (I2 = 0.0%)		36.79	28.17–46.35

Year = publication year of source study; RoB = overall JBI risk-of-bias rating (L = low, M = moderate, H = high). Wilson score confidence intervals computed from the exact numerator/denominator reported in each primary study. Where x = 0 was not applicable, no continuity correction was required

Brugia malayi natural infections across 14,523 Mansonia and Anopheles mosquitoes remained critically low (pooled 0.35%, 95% CI 0.11–1.09%; I² = 91.6%), with Mn. uniformis at 0.02–0.43% (n = 6204), An. donaldi peak with 4.46% (n = 516) [11]. Experimental infections were far higher (pooled 72.43%, 95% CI 42.38–90.37%; I² = 91.5%) with An. campestris at 91.7% and Mn. uniformis reaching 100% [16].

Brugia pahangi, the zoonotic parasite of major concern, showed the highest natural infections in suburban Ar. subalbatus from Kuala Lumpur and Selangor (pooled 7.21%, 95% CI 3.86–13.08%; F = 96%): 14.37% (n = 334) [13], 6.74% (n = 801) [12] and 3.88% (n = 1599) [20]. These three large surveys confirm Ar. subalbatus as a highly competent peri-domestic vector linking animal reservoirs to humans. Experimental infections were substantially higher (pooled 45.00%, 95% CI 25.32–66.38%; I² = 0%); Ae. togoi 50% and Cx. quinquefasciatus 25% [14].

Wuchereria bancrofti showed modest natural rates (pooled 2.07%, 95% CI 0.76–5.51%; I² = 87.5%) with an An. balabacensis at 5.88% (n = 17, Banggi Island) [9] and 1.10% (n = 365) [17], Cx. quinquefasciatus showed 6.71% (n = 164) [9] and An. flavirostris showed 0.90% (n = 1001) [17] across 1,547 mosquitoes. Experimental infection was substantially higher (pooled 36.79%, 95% CI 28.17–46.35%; I² = 0%), reaching 41.7% for An. balabacensis and 34.2% for Ae. togoi [9, 17].

Dirofilaria immitis was reported only from experimental infection studies, where infection in urban vectors was high (pooled 29.05%, 95% CI 11.20–57.06%; I² = 77.8%), with Ae. togoi at 55–69% (n = 51) and Ae. aegypti at 52% (n = 75) [14, 19].

Discussion

This is the first systematic review and meta-analysis which synthesises 42 mosquito–parasite records spanning more than 80 years (1940–2025) and showed variability in mosquito infection rates with human and zoonotic filarial parasites across different regions, time periods and vector species in Malaysia. Although the heterogeneity of study designs and the inclusion of both natural and experimental infections limit the ability to derive precise national estimates of vector infection, the overall evidence supports a profound ecological shift in filariasis transmission in Malaysia. Interestingly, human lymphatic filariasis now shows low endemicity across most settings. In contrast, consistently higher infection rates of B. pahangi and D. immitis in peri-domestic mosquitoes indicate the presence of active zoonotic transmission cycles that largely fall outside the scope of traditional LF elimination efforts.

For B. malayi, the natural infection rates across a total of 15,207 Mansonia and Anopheles mosquitoes remain critically low (pooled 0.17%, 95% CI 0.04–0.53%; I² = 97.80%), despite experimental infection rates exceed 80% in Mn. uniformis (100%) and An. campestris (91.7%). In addition, W. bancrofti natural infection rates in a few mosquito species were generally low (0.90–6.71%) despite large sample sizes, whereas An. balabacensis showed 41.7% infection rates in an experimental setting. In natural settings, such low rates are likely due to wild mosquitoes being exposed to low microfilarial densities, not feeding exclusively on infected human or animal reservoirs and may mount immune responses that limit parasite establishment and development within the vector midgut and thoracic muscles [21, 22]. On the other hand, experimental infections use high microfilarial loads, standardised feeding systems and optimal conditions to maximise uptake, yielding far higher infective rates than those observed in field mosquitoes [23]. This contrast suggests that intrinsic vector competence remains high, but that identified transmission is now reduced following four decades of mass drug administration (MDA) which decreased human microfilaraemia, ecological change and accumulated vector control [21, 24]. Taken together with national data showing low human LF infection rates in most implementing units, these findings reinforce the view that Malaysia has substantially reduced classical brugian and bancroftian filariasis, even if small residual foci persist. However, the high heterogeneity in pooled infection rates (I² > 85% for all parasites) also emphasises that local ecological and methodological differences must be considered when interpreting infection-rate estimates.

Sources of heterogeneity and interpretation of pooled estimates

The substantial heterogeneity observed across all four parasite-specific syntheses (I² > 91.6% for B. malayi, 96% for B. pahangi and 87.5% for W. bancrofti in natural infection; I2 = 77.8% for D. immitis in experimental infection, although several experimental infection subgroups showed I² = 0%) reflects multiple, partially confounded sources of variation rather than a single biological gradient. First, the dataset spans more than 80 years (1941–2022) and the ecology of Malaysian filariasis transmission has changed dramatically over that interval pre-MDA swamp and rural transmission of brugian and bancroftian filariasis in the 1940s–1980s is not directly comparable to post-MDA, peri-domestic, zoonotic transmission in suburban Kuala Lumpur and Selangor in the 2000s–2020s. In addition, diagnostic methods evolved over the same period: historical studies relied exclusively on microscopic dissection, which has lower analytical sensitivity than the PCR and COXI sequencing protocols applied in studies from the 2010s onwards and dissection may also introduce false negatives when larvae are at sub-detectable densities or are confined to tissues not routinely examined. Furthermore, ecological settings differ widely between studies, ranging from freshwater swamps and forest-fringe habitats (where Mansonia spp. and forest Anopheles spp. dominate) to peri-domestic and urban environments (where Ar. subalbatus, Ae. togoi and Cx. quinquefasciatus predominate). Besides that, natural and experimental infection data measure fundamentally different quantities; natural infection rates are joint products of microfilarial exposure, blood-feeding behaviour and intrinsic competence, while experimental infection rates approximate intrinsic competence under standardised, often high-microfilaraemia conditions and combining them in any single estimate is biologically misleading. For these reasons, the pooled proportions presented in Sect.  3.3 should be regarded as descriptive summaries rather than precise prevalence estimates and their 95% confidence intervals should be interpreted cautiously given I2 values consistently above 96%. Therefore, the ranges and patterns described by mosquito–parasite combination and study design were preferred, rather than to rely on a single pooled proportion as a national infection estimate throughout the systematic review.

In contrast, the infection patterns for zoonotic filarial are more worrying, where Brugia pahangi infection rates showed 3.9–14.4% in Ar. subalbatus from suburban Selangor and Kuala Lumpur, together with experimental infection rates of 25–50% in Ae. togoi and Cx. quinquefasciatus, reveal robust peri-domestic transmission cycle involving animals and humans. On the other hand, experimental D. immitis infections in Ae. aegypti, Ae. albopictus, Ae. togoi, Ar. subalbatus and Cx. quinquefasciatus show pooled infection rates around 30%, indicating that common urban mosquitoes can sustain canine heartworm transmission and sporadic human infection.

Although this review provides insights on the vector infection patterns and control gaps on filariasis in Malaysia, it also has limitations. The number of eligible entomological studies is low, many are historical and methods differ substantially across time and location. Several vector–parasite combinations are represented by single small experimental studies, and molecular diagnostics have so far been applied mainly in zoonotic settings, which may bias detection towards those systems. Despite these limitations, this review provides a coherent picture of how mosquito infection rates, diagnostics and control efforts intersect in Malaysia.

Animal reservoirs of Brugia pahangi in Malaysia

A robust understanding of zoonotic B. pahangi transmission in Malaysia requires explicit attention to its principal animal reservoirs. Domestic and stray cats (Felis catus) are the most consistently documented vertebrate reservoir for B. pahangi: Malaysian field and clinical studies have reported microfilaraemia in 9–37% of free-roaming cats sampled in suburban Kuala Lumpur and Selangor [9, 25], and ivermectin-based treatment trials in naturally infected cats confirm both susceptibility and the feasibility of veterinary control [26]. Domestic dogs (Canis familiaris) have also been implicated as secondary reservoirs, with concurrent infections by B. pahangi and D. immitis reported from urban and peri-urban Malaysian dog populations [1, 10]. Beyond companion animals, B. pahangi has been recovered historically from wild leaf–monkeys (Presbytis spp.) and silvered langurs in Peninsular Malaysia and from a range of wild carnivores (civets, leopard cats), suggesting an enzootic sylvatic cycle that overlaps with peri-domestic settings at the forest–suburb interface. Documented human B. pahangi infections in Malaysia remain rare but are well-characterised: the index human case in a Kuala Lumpur suburb [8] and a subsequent case from Selangor [9] both occurred in residents living in close contact with stray and owned cats and within the flight range of Ar. subalbatus breeding sites. The peri-domestic ecology of Ar. subalbatus which oviposits in organically polluted water, including drains, septic seepage, discarded containers and animal-waste pools, places infectious vectors in close, repeated contact with cats, dogs and humans in the same housing compound. Risk factors for human spillover identified or suggested in the Malaysian literature, therefore, include: keeping or feeding stray/free-roaming cats; residence within 50–100 m of stagnant or polluted drainage; gardening and outdoor dusk-evening activity coinciding with Ar. subalbatus biting peaks; and absence of integrated companion–animal antiparasitic prophylaxis. Although seroprevalence surveys of human populations in Malaysian zoonotic B. pahangi foci have not yet been published, the combination of high vector infection rates (6.7–14.4% in Ar. subalbatus), demonstrated competence of Ae. togoi and Cx. quinquefasciatus under experimental exposure and confirmed reservoir competence in domestic animals, supports treating zoonotic brugian filariasis as an active One Health priority for Peninsular Malaysia rather than a residual curiosity.

Diagnostic methods and their influence on the infection-rate evidence

Malaysia’s filariasis vector surveillance has shifted from microscopic dissection to molecular tools, and this methodological transition directly shapes the infection-rate evidence synthesised here. Microscopic dissection, the basis of the older studies in our dataset, is the only technique that directly demonstrates infective L3 larvae and, therefore, the sole stand-alone basis for incriminating a vector [9]; however, it is labour-intensive, low-throughput and prone to false negatives at low larval densities [27], so the natural infection rates in the earliest studies likely underestimate true infection. Molecular methods introduced from 2009 onwards, where molecular xenomonitoring (MX) of pooled samples for post-MDA surveillance [28], COXI-based PCR for species-level discrimination of brugian and bancroftian filariae [9, 29], and the emerging LAMP [30, 31] and droplet digital PCR [32] platforms which offer far greater sensitivity and scalability but cannot, in isolation, confirm transmission-competent larvae or establish vector incrimination [24]. This distinction matters for our synthesis: recent PCR-based zoonotic studies detect parasite DNA (exposure or infection) rather than proven transmission, so molecular-era detection rates are not directly comparable with dissection-based competence data and contribute to the between-study heterogeneity observed. Determinations of vector competence, therefore, require integration of entomological, ecological and epidemiological evidence [24]. In practise, the methods are complementary, where dissection for incrimination, PCR and sequencing for species confirmation, xenomonitoring for programmatic post-MDA surveillance, LAMP for rapid field screening and ddPCR for quantitative validation as summarised in Table 4.

Table 4.

Comparative summary of diagnostic methods for filarial parasites in Malaysian mosquito vectors

Method	Sensitivity	Specificity	Cost/test	Throughput	Strengths	Limitations	Best-suited surveillance scenario
Microscopic dissection	Lower (qualitatively); misses sub-patent infections [27]	Highest when larvae morphologically identifiable	Very low	Low (≤ 30 mosquitoes/skilled worker/day)	Direct visualisation of L1–L3 larvae; only method that demonstrates infective larvae and, therefore, vector incrimination [24]	Labour-intensive; requires trained microscopist; false negatives at low parasite densities; species-level differentiation of brugian and bancroftian L3 morphologically difficult	Vector incrimination studies; gold-standard confirmation of competence
Conventional PCR/COXI sequencing	Higher than dissection; reliably detects sub-patent infections	Very high with sequencing-based confirmation [9]	Moderate	Moderate (96–192 pools/run)	Differentiates closely related filarial species; archivable amplicons; species confirmation by sequencing	Requires laboratory infrastructure; cannot, on its own, confirm presence of infective larvae or vector incrimination [24]	Routine xenomonitoring; species confirmation; post-MDA monitoring
Molecular xenomonitoring (MX, pooled PCR)	High in pooled samples	High when validated primers used	Moderate–high	High (large mosquito pools tested per run)	Designed for very low residual transmission; flags re-emergence after MDA [28, 29]	Cost and reagent supply; assumes representative trapping; cannot distinguish infected from infective mosquitoes without staging	Post-MDA surveillance; verification of interruption; hotspot detection
LAMP (loop-mediated isothermal amplification)	High in optimised assays	Moderate; risk of non-specific amplification [31]	Low–moderate	Moderate (field deployable)	Isothermal, field deployable, no thermocycler required; rapid [30]	Primer design critical; false positives if not optimised; limited multiplexing	Rapid hotspot screening; outbreak response; resource-limited active transmission settings
Droplet digital PCR (ddPCR)	Highest (single-larva detection in pools) [32]	Very high	Very high	Low–moderate (specialised platform)	Absolute quantification; ultra-low detection limit; suitable for pool size optimisation	Expensive instrumentation; requires advanced laboratory and trained operators; not yet routine in Malaysia	Verification studies; pre-validation of MX pool sizes; research-grade confirmation

Limitations: unable to allow vector incrimination

Vector control and the programmatic context in Malaysia

Malaysia’s LF control has evolved through several phases that form the programmatic backdrop to our findings. Early efforts (1960s–1970s) relied on weekly diethylcarbamazine to clear the human microfilarial reservoir [33–36]. From the 1990s, vector-directed measures were added, particularly in Sarawak: ultra-low-volume Malathion fogging for outbreak response [37, 38], pirimiphos–methyl indoor residual spraying [39], and Bacillus sphaericus larviciding against Mansonia and other larvae [40–44]. A national programme delivered five annual rounds of mass drug administration with diethylcarbamazine plus albendazole between 2004 and 2008 at about 80% coverage, although Transmission Assessment Surveys in 2010–2011 revealed residual transmission [45] and insecticide-treated and long-lasting insecticidal nets were scaled up from 2016 as a dual malaria filariasis measure [46–48] (Fig. 6). This trajectory explains the very low contemporary natural infection rates we found in the classical Mansonia and Anopheles vectors, while underscoring that these human and indoor-focused measures do not target the peri-domestic zoonotic B. pahangi and D. immitis cycles that our synthesis identifies as the persistent concern.

Fig. 6.

Timeline of lymphatic filariasis control strategies and intervention in Malaysia

Several factors sustain these vector-control challenges. Vector species exploit highly heterogeneous breeding habitats that is Mansonia in vegetation-rich water, Culex and Anopheles in drains, paddies and ponds, so no single larval-source measure suffices [49]. The exophagic, nocturnal biting of many vectors limits net- and IRS-based control and decades of organophosphate and pyrethroid use have selected for insecticide resistance [50]. Short (7–10 days) developmental cycles allow rapid population recovery after fogging or larviciding [51], while a tropical climate with high rainfall and periodic flooding continually replenishes habitats [52]. Finally, migrant workers from endemic countries can introduce active W. bancrofti and B. malayi infections (seroprevalence up to about 15% in some cohorts), forming cryptic human reservoirs that complicate elimination [26, 53].

Environmental drivers and surveillance implications

Filariasis transmission is increasingly shaped by environmental drivers the included studies were not designed to capture. Warming shortens vector larval development and the parasite’s extrinsic incubation period, with modest warming (1–2 °C) linked to measurable reductions in the extrinsic incubation period and expansion of seasonally suitable transmission windows in tropical Southeast Asia [51, 54]; more frequent intense rainfall replenishes the peri-domestic habitats favoured by Ar. subalbatus, Ae. albopictus and Cx. quinquefasciatus that already sustain zoonotic B. pahangi and D. immitis cycles. Urbanisation similarly expands habitat for Ar. subalbatus and Ae. togoi precisely the species incriminated in our pooled zoonotic estimates, while deforestation has altered the anopheline fauna of Sabah and Sarawak, where An. donaldi replaced An. balabacensis as the dominant vector after habitat modification [55]. The current surveillance system, built around swamp Mansonia and forest Anopheles, is unlikely to capture these shifts, arguing for climate-sensitive, GIS-based entomological surveillance.

Implications for one health filariasis elimination

Sustaining progress will require a One Health approach extending beyond human MDA (Fig. 7). Triple-drug MDA (ivermectin, diethylcarbamazine and albendazole), strengthened community engagement and integrated vector management have advanced elimination, and sterile-insect releases of irradiated male Ae. aegypti have reduced larval indices in urban settings, though rebound must be monitored [53]. Critically, our finding of high B. pahangi rates in peri-domestic Ar. subalbatus and zoonotic D. immitis in urban Aedes points to animal reservoirs that human-focused measures miss: veterinary macrocyclic-lactone (ivermectin) treatment of cats and dogs can help close these animal–mosquito–human pathways [26, 56]. Longer-term environmental measures such as urban redesign to remove peri-domestic breeding sites [57–59], biodiversity-based larval suppression in managed wetlands [60] and targeted removal of the Mansonia oviposition plant Pistia stratiotes rather than destructive swamp drainage [61–63] could complement drug-based control while preserving wetland services [54, 55]. Together these interventions address both residual human LF and the parallel zoonotic cycle our synthesis highlights.

Fig. 7.

One Health framework for zoonotic–filariasis elimination in Malaysia

Prioritised, Malaysia-specific actions

To operationalise this framework, Table 5 ranks candidate interventions by feasibility and expected impact, assigns short-/medium-/long-term timelines, designates lead Malaysian agencies and partners and proposes Malaysia-specific success metrics. Three cross-cutting design choices are particularly important: (i) anchoring zoonotic B. pahangi control, so that companion-animal microfilaraemia is treated as a control target; (ii) using the existing dengue surveillance infrastructure (iDengue, COMBI, Ovitrap network) as the backbone for peri-domestic filariasis xenomonitoring, since the most important filariasis-relevant peri-domestic vectors (Ar. subalbatus, Ae. albopictus) share habitats and surveillance teams with the dengue programme; and (iii) establishing a formal data-sharing mechanism to ensure that veterinary, entomological and human-case data are integrated in near real time. Together these measures aim to bring residual human LF below the WHO verification threshold while actively addressing the parallel zoonotic transmission cycle that currently falls outside human-focused MDA.

Table 5.

Prioritised One Health interventions for filariasis elimination in Malaysia

Intervention	Timeline	Feasibility/impact	Success metric (Malaysia-specific)
Triple-drug MDA (ivermectin + DEC + albendazole) in residual Sabah/Sarawak foci	Short term (0–2 years)	High feasibility/high impact	≥ 80% therapeutic coverage in all 5 remaining endemic sub-districts; reduction of microfilaraemia prevalence to < 1% on Transmission Assessment Survey (TAS)
Veterinary ivermectin/macrocyclic–lactone programmes for free-roaming cats and dogs in zoonotic B. pahangi hotspots (Selangor, Kuala Lumpur)	Short term (0–2 years)	Moderate feasibility/high impact	≥ 70% of stray/owned cats in target catchments treated annually; ≥ 50% reduction in feline microfilaraemia at year 2
Integrated peri-domestic vector control targeting Ar. subalbatus and Ae. togoi (source reduction + biological larviciding) integrated with dengue COMBI activities	Short term (0–2 years)	High feasibility/moderate impact	≥ 30% reduction in Ar. subalbatus larval indices in hotspot enumeration areas; documented co-implementation with dengue Ovitrap surveillance
Molecular xenomonitoring (MX) integrated with the National Dengue Surveillance platform (iDengue)	Medium term (2–5 years)	Moderate feasibility/high impact	MX deployed in all post-MDA evaluation units annually; ≥ 95% of sub-districts with negative MX maintain dossier for WHO verification
Formal veterinary–human data-sharing mechanism (MoH–DVS One Health Filariasis Dashboard)	Medium term (2–5 years)	Moderate feasibility/high impact	Operational dashboard updated quarterly; ≥ 3 published joint MoH-DVS situational reports per year
Community engagement for pet owners and rural/urban residents (responsible cat/dog ownership; bite-avoidance; source reduction)	Short to medium term (0–5 years)	High feasibility/moderate impact	≥ 60% of households in zoonotic hotspots reached by annual outreach; documented behaviour-change KPIs (pet deworming uptake, container-removal rates)
Climate-sensitive surveillance: GIS-based habitat modelling + entomological co-monitoring with meteorological covariates	Medium term (2–5 years)	Moderate feasibility/moderate impact	Annual updated risk map for all 14 states; quarterly entomological reports linked to rainfall/temperature anomalies
Validation of rapid molecular point-of-care diagnostics (LAMP/ddPCR) for filarial xenomonitoring	Long term (5–10 years)	Lower feasibility (R&D-dependent)/high impact	≥ 1 LAMP assay validated and registered for routine Malaysian use; cost per test reduced to <50% of conventional PCR
Sterile insect technique (SIT)/Wolbachia integration with dengue programmes for shared peri-domestic vectors	Long term (5–10 years)	Lower feasibility/moderate impact	Documented reduction of Ae. aegypti/Ae. albopictus indices in pilot sites; assessment of incidental impact on filariasis-relevant peri-domestic culicines

Limitations

Several limitations of this review should be acknowledged, so that the synthesised infection rates and ecological inferences are not overinterpreted. One of the limitations is that the included evidence base is small (n = 12 primary studies) and several mosquito–parasite combinations are represented by single studies with limited sample sizes (for example, n = 1 for Mn. dives experimental infection; n = 4 for Cx. quinquefasciatus experimental B. pahangi infection); confidence intervals around such estimates are necessarily wide and should not be extrapolated. In addition, the dataset spans 1941–2022 and contains substantial temporal heterogeneity (pre-MDA versus post-MDA eras, different vector-control regimes, evolving diagnostics) that cannot be fully adjusted for using the available variables. Third, geographical coverage is uneven: most contemporary zoonotic–filariasis data come from suburban Selangor and Kuala Lumpur, while the interiors of Sabah and Sarawak, where residual W. bancrofti and brugian transmission is most likely to persist remain under-sampled entomologically. Moreover, the combining of natural and experimental infection data within the same parasite-specific synthesis is informative for vector competence but biologically conflates intrinsic susceptibility with field exposure, and the two should not be compared directly. On top of that, individual-level data (vector age structure, blood–meal source, parity, mosquito longevity) were not consistently reported and were, therefore, not extracted, limiting our ability to compute classical entomological inoculation rates or to decompose competence into susceptibility, longevity and biting frequency. Last but not least, the systematic-review framing means that only published articles were synthesised and other Malaysia-based studies; unpublished surveillance datasets held by the government agencies and regional health departments are not represented and may meaningfully shift the picture if subsequently released.

Conclusion

Malaysia’s experience shows that filariasis transmission has evolved from single parasite–vector pair to complex networks of human and zoonotic cycles involving multiple mosquito genera. Classical lymphatic filariasis due to B. malayi and W. bancrofti is now associated with low infection rates in historically important Mansonia spp. and Anopheles spp. vectors, suggesting that mass drug administration, environmental change and vector control have substantially reduced effective transmission, even though vector competence remains high. In contrast, zoonotic B. pahangi and D. immitis show comparatively high infection rates in peri-domestic Armigeres spp., Aedes spp., and Culex spp. mosquitoes from suburban and urban settings, highlighting a growing role for domestic animals and peri-domestic vectors in sustaining filarial transmission. In order to achieve sustainable elimination, both human and veterinary interventions within a comprehensive One Health framework are required. This includes maintaining high-quality MDA in residual human LF foci, strengthening molecular xenomonitoring and GIS-based surveillance originally developed for dengue control and developing ecologically tailored vector management strategies. High resolution mapping targets both swamp-associated Mansonia spp./Anopheles spp. and peri-domestic in Armigeres spp./Aedes spp./Culex spp. populations, supported by real-time outbreak prediction platforms that enable dynamic adaptation to Malaysia’s diverse transmission settings.

Author contributions

VSWY developed the protocol, performed meta-analysis in R, generated forest plots and interpreted heterogeneity. VSWY and LP conducted the literature searches across databases, data extractions, performed abstract screening and drafted the manuscript. JNK and LVL supervised the review process, resolved discrepancies and reviewed and provided critical revisions. All authors read and approved the final manuscript.

Funding

This work was supported by Ministry of Higher Education, Malaysia for niche area research under the Higher Institution Centre of Excellence (HICoE) programme (TIDREC-2023).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.