A Test-and-Not-Treat Strategy for Onchocerciasis in Loa loa –Endemic Areas

1Mass drug administration (MDA) with ivermectin-containing regimens is the main strategy for elimination of lymphatic filariasis and onchocerciasis. Although generally safe, ivermectin distribution has led to severe adverse events (SAEs) in central African countries. More than 500 cases of characteristic post-ivermectin encephalopathy including ~60 fatal case, occurred during MDA and have therefore been reported to the Mectizan Donation Program since 1990. Of note, these neurologic SAEs have occurred exclusively in individuals with peripheral blood Loa loa microfilarial densities >30,000 microfilariae (mf) per milliliter, and are presumed to be related to eosinophil-mediated inflammation around dying mf and/or micro-embolization with subsequent loss of central nervous system vascular integrity.

2Current WHO guidelines allow ivermectin-based MDA to be implemented in areas where onchocerciasis is meso- or hyperendemic because the potential benefits of MDA are felt to outweigh the risk of ivermectin-associated SAEs, although enhanced surveillance for adverse events (AEs) is required. However, areas endemic for loiasis and hypo-endemic for onchocerciasis are spread throughout Central Africa, and remain a serious problem. For these areas, a “Test and (not) Treat” (TaNT) strategy has been proposed, wherein individuals with high L. loa mf loads (at risk for SAEs) are excluded from ivermectin treatment and the remaining population (typically >95%) can be treated safely.

3Successful implementation of the TaNT strategy requires a rapid, point-of-contact, fieldfriendly and highly accurate method to quantify L. loa mf. To this end, a mobile phone-based videomicroscope – the LoaScope (previously CellScope Loa) – was developed. The LoaScope automatically counts L. loa mf in peripheral blood collected in disposable rectangular capillaries without the need for sample processing using a smartphone coupled to a simple optical device (Figure S1 and Movie S1).

4To advance O. volvulus elimination in L. Loa co-endemic countries in Central Africa, we tested the feasibility of this TaNT strategy in the Okola health district in Cameroon, where ivermectin distribution was halted in 1999 after the occurrence of Loa-related encephalopathy. As such, the TaNT strategy allowed the safe reintroduction of ivermectin in all of the communities in this health district without provoking a single SAE.

5The Okola health district (Figure S2) includes 11 health areas where, in 1999 MDA was halted by the Ministry of Public Health following 23 cases of encephalopathy that occured during the first treatment campaign. MDA only resumed in 5/11 areas deemed hyper- or mesoendemic (>20% onchocercal nodule prevalence in adult males). In 2013, nodule surveys in the 6 excluded health areas had prevalences of 6% to 40% consistent with hypo- or meso-endemic onchocerciasis. The entire Okola health district is known to be highly endemic for L. loa.

6The TaNT strategy was implemented in the 92 villages of the 6 health areas untreated since 1999 (Figure S2). The timeline of the TaNT project is depicted in Figure S3 and the process is described in the Supplementary Methods. All individuals aged >=5 years were invited to participate.

7The TaNT process consisted of registration of consenting (or assenting) individuals >=5 years of age, LoaScope quantification of L. loa microfilarial density, treatment of eligible individuals with ivermectin (150μg/kg) and surveillance for AEs. Non-pregnant subjects excluded from ivermectin distribution because of high L. loa mf counts were given albendazole (400 mg) for intestinal deworming. Self-declared pregnant women were not treated with ivermectin or albendazole, but were offered iron and folic acid tablets. Each participant was given a card (Figure S4) with their L. loa mf count, the treatment received and a contact phone number for questions and/or reporting of AEs.

8The use of the LoaScope and its performance have been described previously. A threshold of 26,000 mf per milliliter was initially selected for ivermectin treatment, based on the lower 95% confidence interval around the 30,000 mf per milliliter threshold below which no neurologic SAEs were observed in prior studies and the calculated false negativity rate of 1 in 10 million (<0.00001%). Two weeks after the study start, a case of conjunctival hemorrhage, similar to those described previously, occurred in a subject with a LoaScope L. loa mf count of 24,599 mf per milliliter. For potential safety reasons, the exclusion threshold of the LoaScope was decreased to 20,000 mf per milliliter for the remainder of the trial.

9Calibrated (50 μL) thick smears were performed as a backup for samples unable to be analyzed with the LoaScope, to identify and quantify Mansonella perstans mf, and to corroborate the accuracy of the LoaScope. Smears were read by 2 different microscopists blinded to the LoaScope results. Dried blood spots collected on filter paper were archived and stored at -80°C.

10Ov16 IgG4 antibodies positivity from eluted single blood spots (10 μl equivalent) was determined the SD Bioline Onchocerciasis IgG4 Rapid Test.

11Results were read and recorded at 24 hours.

12Monitoring for AEs was performed by two surveillance teams, each composed of a physician and a driver, with the assistance of selected community members (pre-Community Drug Distributors (pre-CDDs)) and local nurses. The surveillance teams visited each village on days 1, 2, 3 and 6 post-treatment, examined all individuals complaining of AEs and provided symptomatic treatment if indicated. In addition, the team toured the entire community by car to identify additional individuals with AEs. All AEs were recorded using a standardized form (Figure S5). A Karnofsky performance score index was assigned to each patient examined. Clinical management was based on reference guidelines.

13Medians and interquartile ranges were used as measurements of central tendency. Associations between individual factors (gender, age, L. Loa mf density (assessed by LoaScope), presence of Ov16 IgG4, presence of M. perstans mf (assessed by calibrated blood smear microscopy) and the occurrence of AEs were assessed using multivariable logistic regression.

14The logistic regression coefficients were used to calculate population attributable fractions. The calibrated thick smear was used as the reference test for assessment of the specificity and negative predictive value of the LoaScope.

15This study was authorized by the National Ethics Committee of Cameroon (ethical clearance n° 2013/11/370/L/CNERSH/SP) and approved by the Division of Operational Research at the Ministry of Health (Administrative authorization n° D30- 571/L/MINSANTE/SG/DROS/CRSPE/BBM). All volunteers provided written signed consent (or parental consent in the case of minors) before undergoing blood sampling and again before receiving treatment.

16Authors’ contributions to this study are as follows: JK, SDP, CDM, ADK, TBN and MB designed the study; JK, SDP, CBC, MHB, MVDA CDM, HCND, RGK, GRN, PN, JBTM, SW, DAF, ADK, TBN and MB gathered the data; SDP and CBC analyzed the data; TNB and MB vouch for the data and the analysis; MHB, MVDA, DAF provided diagnostic technology development and support; JK, SDP, CDM, HCND, WAS, DAF, ADK, TBN and MB wrote the paper; and JK, SDP, CDM, DAF, ADK, TBN and MB decided to publish the paper. SDP wrote the first draft of the manuscript.

17A total of 16,259 individuals were examined during the TaNT process (Figure 1). The median age of the examined populations ranged from 17 to 26 years in the different health areas and the sex distribution was relatively equal (48% male). The prevalence of Ov16 IgG4 antibody in the six health areas varied from 15.3% to 29.9%. The prevalence of L. loa microfilaremia varied from 15.3% to 22.8%, and the proportion of individuals with more than 20,000 Loa mf per milliliter as determined by LoaScope ranged from 1.3% in the Ngoya health area to 2.4% in the Nlong and Ekekam III health areas (Table 1, Table S2).

18Between 50 and 100 participants were typically examined per village per day. The mean time from finger prick to LoaScope result was 2-3 minutes. The LoaScope results were immediately available for 16,099/16,259 individuals (99%) and were delayed for 160 individuals (1%) because of technical problems requiring determination of the mf count by calibrated thick smear. Ivermectin was administered to 15,522 individuals (95.5%) with mf levels below the established threshold.

19Seven hundred and thirty seven (4.5%) subjects were excluded from ivermectin therapy. Of these, 340 (2.1%) were excluded because of a L. loa density above the risk-threshold, 228 (1.4%) because of poor health (signs or symptoms consistent with a serious acute or chronic concomitant illness) or inebriation, and 169 (1%) because of pregnancy or breastfeeding. The proportion of excluded individuals per village varied from 0% to 15.1% (Figure S6). All excluded individuals (except pregnant women), were treated with albendazole (400 mg). The median treatment coverage in the district was 55% of the total population (interquartile range between villages: 42.9–64.1%), and 64% of the targeted population.

20The prevalence of O. volvulus-specific antibody (Ov16 IgG4) was 22.0% in individuals who received ivermectin, 25.4% in those excluded for pregnancy or illness, and 33.5% in those with a L. loa density above the risk-threshold. Thus, individuals who were not treated because of Loa microfilaremia and who were potentially infected with O. volvulus represented only 0.7% of the examined population.

21Among the 15,522 individuals treated with ivermectin, 934 (6%) had documented AEs. The incidence of AEs decreased slightly from 6.6% (464/7,065) to 5.6% (470/8,457) (p<0.0001) after reducing the exclusion threshold from 26,000 to 20,000 mf per milliliter. Dermatologic manifestations were most common, followed by systemic and rheumatologic manifestations (Table 2). Eight hundred and sixty-nine people (93%) had a Karnofsky score of 90, and 65 (7%) had a score of 80. All AEs resolved within one week without treatment or with basic supportive therapy (anti-histamines, non-steroidal anti-inflammatory drugs, or acetaminophen).

22Both L. loa microfilaremia and the presence of Ov16-specific IgG4 were assessed in 888 of the 934 individuals who developed an AE. Among these, 43.2% had neither L. loa mf nor Ov16 IgG4, 22.3% had only L. loa mf, 23.9% had only Ov16 IgG4, and 10.6% had both L. loa mf and Ov16 IgG4. Multivariable regression indicates that AEs were significantly more frequent in older individuals, females, and individuals with either L. loa mf or Ov16 IgG4 (Figure 2). The risk of AEs associated with presence of Ov16 IgG4 was similar to that associated with harboring 1-8000 Loa mf per milliliter (Odds ratio (OR)=1.61 and 1.71, respectively) and was about half the risk associated with harboring 8000-20,000 Loa mf per milliliter (OR=3.00). The risk of AEs associated with both L. loa microfilaremia and O. volvulus IgG4 was similarly increased in persons harboring 1-8000 L. loa mf per milliliter (OR=2.47) and in those harboring 8000-20,000 L. loa mf per milliliter. Population attributable fractions of AEs for L. loa mf density of 1-8000 mf per milliliter, 8000-20,000 mf per milliliter and Ov16 IgG4 were 8.0%, 8.3% and 12.2%, respectively.

23Figure 3 shows that the distributions of L. loa mf density in the population using the LoaScope and thick smear microscopy were similar. The specificity and negative predictive values of the LoaScope to identify individuals with mf counts below 20,000 mf per milliliter (as assessed by microscopy) were 99.7% (95% confidence interval: 99.6 – 99.8) and 99.7% (99.6 – 99.7), respectively.

24Extension of ivermectin-based MDA to areas hypoendemic for onchocerciasis and coendemic for loiasis remains a significant obstacle to the success of onchocerciasis elimination programs in Africa. In the current study, a LoaScope-based TaNT strategy was used to safely treat more than 15,000 individuals with ivermectin in such an area. Although there was initial reticence to participate in some villages because of the memory of the SAEs (including deaths) that occurred in 1999, 16,259 of the 22,842 individuals aged >= 5 years old recorded during the initial census (71.1%) participated in the TaNT campaign. This suggests that TaNT is an acceptable strategy even in populations with a history of previous ivermectin-related SAEs. Though not formally assessed, it is likely that fear of SAEs was the main reason for non-participation.

25During the first MDA campaign conducted in 1999 in Okola, 23 cases of neurological SAEs, including three fatalities, were recorded among the 6,000 individuals who received ivermectin before MDA was stopped. The incidences of post-ivermectin neurological SAEs and deaths were therefore 38/10,000 and 5/10,000, respectively. Extrapolating these data to the population enrolled in the present study, a minimum of 62 cases of neurological SAEs and 8 deaths were theoretically prevented by TaNT.

26Although some individuals (6%) complained of ivermectin-associated AEs during the TaNT campaign, the proportion was lower than that typically observed after ivermectin MDA for onchocerciasis in areas not endemic for loiasis: 13.1% in south-east Nigeria, 12% and 20% in northern Cameroon, and 21.4% in eastern Sudan. It was also much lower than the 26.3% recorded in a neighboring Loa-endemic area of central Cameroon. The most likely explanation for the lower frequency of AEs recorded in the Okola district is that onchocerciasis is hypo- and mesoendemic in this area.

27The LoaScope operators underwent a 1-hour training session 2 weeks before the field operations. This training was sufficient for the entire study, and the teams noted the ease of use and reliability of the device despite daily use and demanding field conditions. Because L. loa mf are diurnally periodic, LoaScope examinations (and treatment) started at 10 am and ended at 4 pm. During this TaNT campaign, up to 162 individuals were examined per village per day.

28Whereas the present study clearly shows that the TaNT procedure is safe and feasible at a district level, moving TaNT from the operational research arena to Central African-wide implementation will depend on a number of factors, including greater reliance on local personnel for the census and post-treatment surveillance and smaller teams (a “tester” and a “treater”) for the TaNT process itself. If only a single TaNT round were needed, this would have a major impact on the applicability of this approach at a larger scale. Since ivermectin has marked microfilaricidal and probable embryostatic activity against L. loa, marked and sustained reduction in L. loa mf density is expected for one year after the TaNT campaign. In fact, in a neighboring district, the average reduction in L. loa microfilarial load was >74% one year after a single dose of ivermectin, and no individual with a pre-treatment density <30,000 mf per milliliter had a count above this level after 12 months. Thus, it seems likely that a single community-wide round of TaNT will be necessary, with individual testing in subsequent years restricted to previously ivermectin-untreated individuals. This hypothesis, as well as operationality, performance and cost of a TaNT conducted by 3-member teams will be assessed in late 2017.

29Given the low percentage (2.4%) of the total population excluded from ivermectin treatment and the proposed implementation of TaNT in areas hypo- and meso-endemic for onchocerciasis, it is unlikely that excluded individuals will be a significant reservoir of O. volvulus microfilariae at the community level. Nevertheless, some excluded individuals are likely to be infected with O. volvulus and, for ethical reasons, should be treated with effective and safe drug regimens, particularly in the setting of clinical manifestations of onchocerciasis. Although a 4-6-week course of doxycycline, a regimen known to be macrofilaricidal for O. volvulus but not L. loa24, is impractical at the community level, it could be used safely in this context.

30In summary, this TaNT strategy based on a novel and scalable point-of-contact tool that allows rapid identification (and exclusion from ivermectin-based treatment) of individuals at risk of Loa-related SAEs has enabled district-level community treatment of onchocerciasis. Though this TaNT strategy was motivated by the need to tackle hypoendemic onchocerciasis in Central Africa, it could also be considered for other foci co-endemic for onchocerciasis and loiasis. As many (but not all) meso-hyperendemic areas are already covered by CDTI, TaNT would target ivermectinnaïve individuals and systematic non-compliers.