A total of 4,330 fecal samples from 1,839 refugees underwent DNA extraction and multiparallel qPCR. After excluding 4 participants positive for hookworm at T1 who did not receive albendazole, the number of participants with T1-T2 paired samples totaled 1,548; we had samples from all 3 time points for 517 participants. The geometric mean time between T1 and T2 sample collections was 188 (range 48–1,013) days and between T2 and T3 was 46.2 (range 14–413) days. Baseline hookworm infection (any type) in this population was high, ranging from 25.9%–32.8% depending on the camp of residence (Table 2).

Detection of A. ceylanicum

Figure 2

Figure 2. Partial internal transcribed spacer 2 (ITS2) sequences of Ancylostoma duodenale and A. ceylanicum hookworms. Boxes indicate the location of forward and reverse primer binding; gray shading indicates the location of probe...

When a highly sensitive primer-probe set specific to the A. duodenale genome (Ad2) was used on samples positive for Ancylostoma DNA by qPCR with primer-probe set Ad1, none was positive. Because of concern that Ad1 might enable the cross-amplification of other Ancylostoma spp., we performed seminested PCR and MvaI and Psp1406I digestion with 7 samples positive by Ad1 but negative by Ad2. Results from restriction fragment length polymorphism PCR indicated the presence of A. ceylanicum hookworm (Technical Appendix Figure); these results were further confirmed by sequencing (98% identity to A. ceylanicum ribosomal sequence [GenBank accession no. LC036567]). The internal transcribed spacer 2 regions of the A. duodenale and A. ceylanicum genomes, which the Ad1 primer-probe set aligned with, are identical (Figure 2). Cross-species identification with this primer-probe set has been previously predicted (23), although not previously demonstrated in the literature.

We then determined the prevalence of A. ceylanicum hookworm among the refugee population using a primer-probe set specific to a repetitive DNA element in the A. ceylanicum genome (Ac; Table 1). When using Ac, the total number of A. ceylanicum–positive samples increased from 106 (using the Ad1 set) to 124 (using the Ac set). We tested or retested these samples (n = 124) by qPCR using the Ad2 primer-probe set, and 0% were positive. All samples positive by qPCR with Ad1 were positive by qPCR with Ac.

Response to Treatment

Figure 3

Figure 3. Baseline prevalence of hookworm infections in 1,839 US-bound Myanmar refugees at 3 camps along the Myanmar–Thailand border, by age group, Thailand, 2012–2015. A) Ancylostoma ceylanicum hookworm. B) Necator americanus hookworm.

Baseline prevalence of N. americanus (26.3%) hookworm was higher than that of A. ceylanicum (5.3%) hookworm among all participants (n = 1,839); likewise, prevalence of N. americanus (25.4%) species was higher than that of A. ceylanicum (5.4%) species among all participants who gave paired T1-T2 fecal samples (n = 1,548) (Table 3). Refugees in their sixth (50–59 years) and seventh (60–69 years) decades of life had the highest N. americanus hookworm prevalence (>40% positive), and refugees in their third decade of life had the highest A. ceylanicum hookworm prevalence (9%) (Figure 3).

A. ceylanicum infection had a higher cure rate than did N. americanus infection (Table 3); 92.8% of A. ceylanicum hookworm–infected refugees were cured by T2, despite the relatively long time that elapsed between T1 and T2. Of the samples paired for T2 and T3, all 7 A. ceylanicum hookworm–infected refugees at T2 were cured by T3. At T2, the N. americanus infection cure rate was 69%, and after the second administration of albendazole at T3, 42% (21/50) were cured, resulting in 29 participants with persistent N. americanus infection at resettlement in the United States. Combining cure rates across all time points, the overall A. ceylanicum hookworm cure rate was 93.3% (84/90), higher than that for N. americanus hookworm (65.9%, 292/443; p<0.001).

Figure 4

Figure 4. Change in relative quantities of Ancylostoma ceylanicum and Necator americanus hookworm genomic DNA in fecal samples from US-bound Myanmar refugees at 3 camps along the Myanmar–Thailand border after treatments with albendazole...

In total, 151 refugees had persistent N. americanus infections, and 6 refugees had persistent A. ceylanicum infections. At T2, the hookworm genomic DNA relative quantity in fecal samples of those with persistent N. americanus infections (n = 122) decreased significantly (p<0.001) (Figure 4). Of the 6 persons with persistent A. ceylanicum infections at T2, 4 (67%) had a decrease and 2 (33%) an increase in hookworm genomic DNA in their fecal samples. Those infected with N. americanus hookworm at T2 and T3 (n = 29) had no significant decrease in parasite genomic DNA relative quantity at T3, and 26 refugees were positive for N. americanus hookworm at all 3 time points.

We detected a small number of newly acquired hookworm infections. Of the 1,548 participants with T1-T2 paired samples, we detected 15 (≈1%) new A. ceylanicum infections and 36 (2.3%) new N. americanus infections (Table 4). Of the 517 participants with T2-T3 paired samples, we detected 3 (0.58%) new A. ceylanicum infections and 13 (2.5%) new N. americanus infections.

Blood Cell Concentration Differences Between N. americanus and A. ceylanicum Infections

Figure 5

Figure 5. Eosinophil and hemoglobin concentrations in US-bound Myanmar refugees at 3 camps along the Myanmar–Thailand border who were monoinfected with Ancylostoma ceylanicum (n = 24) or Necator americanus (n = 143) hookworm...

We further evaluated refugees with either N. americanus (n = 143) or A. ceylanicum (n = 24) hookworm monoinfections; participants co-infected with both hookworms or other soil-transmitted helminths (i.e., S. stercoralis, A. lumbricoides, and T. trichiura roundworms) were excluded. Peripheral blood eosinophil concentrations were significantly higher (p<0.001) in those with A. ceylanicum monoinfections (geometric mean 8.49 × 10⁸ cells/L, 95% CI 5.98–12.04 × 10⁸ cells/L) than those with N. americanus monoinfections (geometric mean 3.44 × 10⁸ cells/L, 95% CI 2.92–4.05 × 10⁸ cells/L) (Figure 5, panel A). The hemoglobin levels did not differ between those with only A. ceylanicum and those with only N. americanus infections (Figure 5, panel B).

A. ceylanicum hookworm is increasingly being recognized as a pathogen in humans, particularly in Southeast Asia (32). A. ceylanicum is the only hookworm species known to achieve patency in both humans and other animals (e.g., dogs and cats) (33). In this group of US-bound refugees from Myanmar, N. americanus infection was the most prevalent hookworm infection at all 3 time points tested (25.4% at baseline); A. ceylanicum was the only other hookworm species found, with a baseline prevalence of 5.4%. Because A. ceylanicum infection has been described in both Thailand (7,10) and Myanmar (11,34), whether these persons were infected in their home country or in the camps in Thailand is unknown. Camp allocation seemed to have an effect on infection status for both hookworms; camps 2 and 3 imparted a higher risk for A. ceylanicum hookworm acquisition, and camp 1 had a higher rate of N. americanus infection. Evidence of newly acquired infections for both hookworms at T2 and T3 (Table 4) indicates that active transmission of both species was ongoing at these 3 camps. However, whether the majority of hookworm infections at baseline were acquired prior to entry or while residing in the camps is unknown. Differential infection rates at the 3 camps might reflect environmental and hygiene conditions in the camps, the historical exposure of the persons at these camps, or other factors. Of note, a risk factor for infection with either A. ceylanicum or N. americanus hookworm was infection with the other hookworm (Table 5). This finding contrasts with previous surveys showing that A. ceylanicum and N. americanus co-infections are rare (26,32). Although A. ceylanicum infection by the fecal-oral route is thought to be possible (33), walking barefoot has been shown to be a major risk factor for A. ceylanicum and N. americanus hookworm infections (35). Thus, the high propensity for co-infection in the population we evaluated suggests a similar mode of transmission, namely transdermal penetration, for both hookworms.

Despite possible co-transmission, those infected with A. ceylanicum hookworm had slightly different risk factors than those infected with N. americanus hookworm. A. ceylanicum hookworm prevalence peaked in the third decade of life, compared with N. americanus prevalence, which peaked in the sixth and seventh decades. However, N. americanus prevalence remained high for many decades of life (20–69 years of age).

T. trichiura roundworm, Entamoeba histolytica ameba, and A. lumbricoides roundworm infection as risk factors for N. americanus hookworm co-infection reflects the high prevalence of parasitic infections in this population (21), in a setting with inadequate sanitary infrastructure despite improvement efforts (36). Why infections with these pathogens but not S. stercoralis roundworm (acquired similarly to hookworm) or Giardia duodenalis protozoa (acquired similarly to E. histolytica ameba) put refugees at risk for N. americanus infection deserves further study.

Female participants were less likely to acquire both N. americanus and A. ceylanicum hookworms, a finding that might reflect differential exposures, differences in immunity (37,38), differences in albendazole metabolism (39), or a combination of these factors. Although eosinophilia has been previously reported in experimental human infections with A. ceylanicum hookworm (33) and in A. ceylanicum case reports (34,40), a more striking eosinophilia was seen among those with A. ceylanicum infections than those with N. americanus infections. This observation supports the idea that, unlike the hookworms that only infect humans, the zoonotic A. ceylanicum parasite might be less able to downregulate the host’s IgE-mediated response to infection (41). No difference in hemoglobin levels was found between those with either hookworm species. Anemia is typically seen in A. duodenale infection and is less commonly associated with N. americanus infection (3); for A. ceylanicum infection, data on anemia are scant.

The treatment for N. americanus infection was only modestly effective (cure rate 42%–69%). It has been suggested that deworming might have enabled the zoonotic A. ceylanicum hookworm to fill a niche left by a decrease in anthropophilic hookworms (32). That A. ceylanicum infection was largely cured after single courses of treatment with albendazole (cure rate 92.8%–100%) is reassuring, although follow-up time periods in this evaluation were more varied than in most controlled studies specifically assessing response to treatment. Still, reinfection and newly acquired infection rates (0.58%–0.97% for A. ceylanicum and 2.3%–2.5% for N. americanus) were modest after deworming, compared with previous projections suggesting reinfection rates as high as 30% (14) after 3 months. With the average time between T1 and T2 exceeding 6 months (and the wide time range of 1–33 months), this population experienced a lower reinfection rate than has been suggested for endemic areas.

Persons who were infected with N. americanus hookworm at all 3 time points represent a group that might have never cleared the infection or might have cleared infection but were subsequently reinfected. However, the time between sample collections was not statistically different between those who were persistently positive and those who cleared infection after a single treatment (E.M. O’Connell, unpub. data). Reports of benzimidazole resistance are increasing in the literature on veterinary medicine; resistance in canine Ancylostoma caninum hookworm (15) and phylogenetically similar bovine intestinal nematodes (17) was associated with SNPs in the β-tubulin gene, particularly in codon 200 but also in codon 167 and, rarely, in codon 198. Likewise, other studies have shown low cure rates after benzimidazole administration in the setting of N. americanus infection, and research suggests that resistance is emerging (42). One group found that albendazole administration exerts selective pressure on T. trichiura codon 200 in Kenya and Haiti (20). The same group found that, in pooled N. americanus eggs from Haiti, the allele containing the resistant codon 200 had a mean allelic frequency of 36% (43). A frequency of 0% was found for the homozygous resistant genotype in hookworm eggs (species not identified) from Haiti and Panama (20), and although the authors reported a 2.3% frequency of the homozygous resistant genotype in hookworm eggs from Kenya, the frequency of the resistant genotype after treatment did not increase (20), raising questions about the significance of this allele. Our examination of codons 167 and 200 in samples persistently positive for N. americanus hookworm across all 3 time points, and of codon 200 in those with high levels of N. americanus DNA in fecal samples at any time point, revealed only homozygous wild-type β-tubulin genes.

Several possibilities might explain why β-tubulin mutations were not found to account for persistent N. americanus infections. First, the lesser-known codon 198 or another codon within the β-tubulin gene might be responsible for resistance in the hookworms infecting these refugees. Second, whereas amplification rates for our qPCR assay were as high as or higher than those in most other reports, up to 54% of samples positive for N. americanus DNA at T3 did not amplify adequately to determine genotype. Therefore, mutations in SNPs at either codon 167 or 200 could have been missed. Last, β-tubulin might not be the only gene involved in drug resistance and responsible for low cure rates in N. americanus. In this population, 88.5% (23 of 26) of those persistently positive for N. americanus at all 3 time points were male. The peak concentration in serum and area under the serum concentration time curve for albendazole sulfoxide and albendazole sulfone (the main active metabolites of albendazole) have been found to be higher in female than in male volunteers (39). Also, male refugees were possibly more likely than female refugees to quickly reacquire infection due to differences in environmental exposures.

One lesson from this evaluation was that unbiased surveys are necessary before pursuing highly specific molecular techniques when profiling organisms that infect a population, particularly when considering emerging infectious diseases. Although the use of the highly repetitive genomic sequence for A. duodenale hookworm is superior in sensitivity and specificity than the internal transcribed spacer 2 region (44), targeting the more conserved genomic region ultimately enabled detection of an unexpected organism, A. ceylanicum hookworm, which otherwise would have been missed in this population.

This project had several limitations. Control and nontreatment groups were absent. Participation was completely voluntary. The number of fecal sample collections for the third time point dropped off substantially because of logistical issues relating to collecting and shipping samples from the various US states after resettlement. Unlike other studies focused on A. ceylanicum hookworm, this evaluation did not include surveying the local cat and dog populations to establish potential reservoirs, which might be a worthwhile future research direction.

In summary, this cohort of US-bound refugees living in 3 camps in Thailand on the Myanmar–Thailand border was found to have a high prevalence of N. americanus hookworm with suboptimal cure rates after albendazole administration that do not seem to be attributable to mutations in the β-tubulin gene at codons 200 or 167. In addition, A. ceylanicum hookworm was the only other hookworm species identified in this population. A. ceylanicum infection had a much higher cure rate after a single course of albendazole, and those with A. ceylanicum monoinfection had a similar hemoglobin level as those with N. americanus monoinfection. Future mapping efforts of soil-transmitted helminths should take into account the emergence of A. ceylanicum hookworm infection in humans to further understand its distribution across the world. The recognition of the increased importance of zoonotic A. ceylanicum hookworm over that of A. duodenale hookworm in some populations raises epidemiologic questions about transmission dynamics and the differential effect on local health of these 2 species.