Infections caused by Haemophilus ducreyi: one organism, two stories

SUMMARY

Chancroid, a sexually transmitted infection caused by Haemophilus ducreyi, is characterized by painful genital ulcers (GU) and inguinal lymphadenitis. H. ducreyi was recently described as a major cause of non-sexually transmitted cutaneous ulcers (CU) on the lower legs in children in yaws-endemic regions. This review explores the relationship between CU and GU strains of H. ducreyi; their clinical presentation, diagnosis, epidemiology, and treatment; and how findings from a human challenge model relate to GU and CU. We contrast the decline of GU with the persistence of CU caused by H. ducreyi. Factors such as transmission dynamics, control, and elimination efforts are discussed. Syndromic management and targeted treatment of sex workers can eradicate chancroid, while skin colonization by CU strains and environmental factors may necessitate topical treatments or vaccination for CU eradication. Efforts should focus on identifying additional reservoirs of CU strains, improving hygiene, and eliminating asymptomatic colonization to eradicate this painful infection in children.

INTRODUCTION

Chancroid is a sexually transmitted infection (STI) caused by Haemophilus ducreyi, which primarily infects skin and mucosal surfaces. Chancroid is characterized by painful genital ulcers (GU) and inguinal lymphadenitis and facilitates the sexual transmission and acquisition of the human immunodeficiency virus (HIV-1) (1–3). H. ducreyi was recently found to be a major cause of non-sexually transmitted exudative cutaneous ulcers (CU) in children who live in yaws-endemic regions of the South Pacific and Africa (3, 4).

In this article, we discuss the relationship between CU and GU strains of H. ducreyi; the clinical presentation, diagnosis, treatment, and epidemiology of chancroid and CU; and the pathogenesis of H. ducreyi defined in a controlled human infection model. Given the decline of H. ducreyi as a cause of GU and the persistence of H. ducreyi as a cause of CU, we explore reasons behind potential differences in the control and elimination efforts for these two conditions, including the complex dynamics of transmission, the availability of effective treatment options, vaccine candidates, and the impact of environmental and social factors. By analyzing these factors, we aim to shed light on the challenges posed by H. ducreyi as a cause of CU and identify potential strategies for its control.

SEARCH STRATEGY

To provide an overview of Haemophilus ducreyi, we utilized the following PubMed search strategy: (“Haemophilus ducreyi” OR “ducreyi infections” OR “chancroid” OR “cutaneous ulcers”) AND (“pathogenesis” OR “epidemiology” OR “diagnosis” OR “treatment” OR “antimicrobial resistance” OR “prevention”). This search technique combined several Haemophilus ducreyi keywords with several infection-related keywords, such as pathogenesis, epidemiology, diagnosis, treatment, resistance to antibiotics, and prevention.

HISTORICAL CONTEXT AND MILESTONES

Chancroid is believed to have existed in human populations since the times of the ancient Greeks. The historical background and controversies surrounding the disease, as well as the isolation of its causative agent, H. ducreyi, have been extensively covered in several comprehensive reviews (5, 6). Although chancroid (soft chancre) was clinically distinguished from syphilis (hard chancre) in the 1850s, H. ducreyi was not isolated in culture until 1900. However, the lack of a selective media for routine isolation of H. ducreyi from clinical specimens posed a significant challenge for chancroid research. A major advance occurred in the late 1970s when Hammond et al. developed selective media using commercially available reagents (7). This enabled the isolation of H. ducreyi from clinical specimens, paving the way for epidemiological studies that demonstrated the association between chancroid and the acquisition of HIV-1. Subsequently, these findings prompted extensive research on pathogenesis, molecular diagnostics, therapy, and eradication strategies for chancroid.

More recent milestones in the history of chancroid were the recognition that the disease could be eradicated by providing antimicrobial prophylaxis to sex workers, who serve as the clinical reservoir for H. ducreyi, and the widespread application of syndromic management for GU by the World Health Organization in 2001 (1, 8). Syndromic management initially included empiric treatment of GU for both chancroid and syphilis with no diagnostic testing and made it easier to manage chancroid in environments with limited resources. By shortening the duration of infection, antimicrobial prophylaxis of sex workers and syndromic management decreased transmission and led to dramatic declines in the prevalence of chancroid during the past two decades (4, 8), such that current syndromic management of GU no longer includes treatment for H. ducreyi.

H. ducreyi was long thought to be exclusively sexually transmitted, and infections in children were attributed to sexual abuse. Between 2006 and 2010, several case reports of children with no history of sexual abuse and adults who visited or emigrated from yaws-endemic areas and developed painful CU on their lower legs from which H. ducreyi was isolated were published (9–12), confirming an initial case report published in 1989 (13). This led to the application of PCR-based tests for H. ducreyi on CU specimens obtained during subsequent yaws eradication campaigns; H. ducreyi was identified as a major cause of CU in children who live in equatorial regions of the South Pacific and Africa (3, 4).

TAXONOMY, STRAIN CLASSIFICATION, AND COMPARATIVE GENOMICS OF CU AND GU STRAINS

H. ducreyi is a Gram-negative coccobacillus that is not a true Haemophilus species. Within the Pasteurellaceae, H. ducreyi is grouped in a lineage with Aggregatibacter (formerly Actinobacillus) pleuropneumoniae and Mannheimia haemolytica; H. ducreyi diverged from these animal respiratory pathogens to occupy its niche in the human epithelium (14).

There are two circulating classes of H. ducreyi, which express different variants of several outer membrane proteins (15) and different proteomes (16). Compared to Class I GU strains, Class II GU strains express a more truncated lipooligosaccharide, grow more slowly on agar plates, and may be underreported since their growth is inhibited by vancomycin, which is included in selective media used to isolate the organism (15). Whole-genome phylogenetic analyses of GU strains with a worldwide distribution isolated between 1954 and 1995 and CU strains isolated from the South Pacific Islands and Africa between 2006 and 2015 showed that Class I and Class II GU strains have sufficient genomic conservation to be considered a single species but form two distinct clades that diverged from each other ~1.9 million years ago (17, 18). The CU strains diverged from multiple lineages of Class I and Class II GU strains during the past 180,000 years and then diversified from each other during the past 27,000 years (17, 18). A limitation of this analysis is that, due to syndromic management of GU, CU strains could not be compared to contemporaneous GU strains, which are unavailable (17, 18). Direct whole-genome sequencing of H. ducreyi DNA-positive CU samples from Ghana and the Solomon Islands confirmed that CU strains diverged from both classes of GU strains (19). A single locus typing system developed from whole-genome sequences showed that multiple Class I and Class II CU strains circulate simultaneously in yaws-endemic communities, that co-infections with both classes are common, and that strain composition is not affected by antibiotic pressure over time, consistent with asymptomatic colonization of intact skin by CU strains and other environmental factors (20, 21). In addition, CU associated with Class II strains tend to be longer lasting than CU associated with Class I strains, with 9.3% of Class II ulcers having a self-reported duration >6 months (21). As this typing system was not available when chancroid was common, similar studies have not been done on GU strains.

CLINICAL FEATURES OF GU AND CU

To cause GU, H. ducreyi is thought to enter the host through breaks in the epithelium that occur during intercourse (5). Erythematous papules form at entry sites within hours to days and evolve into pustules in 2–3 days. After ~2 weeks, the pustules ulcerate, and patients typically have one to four ulcers. Patients usually do not seek medical attention until they have had ulcers for 1–3 weeks, ~3–5 weeks after inoculation (5, 7, 22). Natural ulcers classically are described as very painful and nonindurated with ragged edges (5). The ulcer may be covered by a purulent exudate and bleeds when scraped. However, the classical presentation of chancroid occurs in a minority of patients (23), and chancroid is usually clinically indistinguishable from syphilis and genital herpes. Lesions in men are usually localized to the external and/or internal surfaces of the foreskin and to the coronal sulcus or penile shaft (Fig. 1A) (5). Most lesions in women are at the vaginal entrance (Fig. 1B) (5). In women, asymptomatic carriage detected by PCR is rare (24), but infected women may have internal vaginal and cervical ulcers that are painless. In both men and women, extragenital skin lesions may occur in nearby areas such as the thighs and buttocks and are thought to be due to autoinoculation (5, 7).

Fig 1.

Chancroid. (A) Chancroidal ulcers on the penile shaft and foreskin. (B) Multiple vulvar ulcers. (C) Ulcer on the prepuce with left inguinal lymphadenitis. Reproduced from S. A. Morse, et al. Atlas of sexually transmitted diseases and AIDS, 4th ed. Elsevier Ltd, London, UK (25).

A total of 10–40% of patients with chancroid have suppurative inguinal lymphadenopathy or buboes (Fig. 1C). H. ducreyi is likely trafficked to lymph nodes by dendritic cells that ingest the organism but do not kill it, and H. ducreyi can be recovered from buboes (5, 26, 27). In vitro, H. ducreyi dies at temperatures above 35°C, which likely explains why H. ducreyi does not cause bacteremia (28).

To cause CU, H. ducreyi likely enters the skin of an asymptomatically colonized child via minor traumatic wounds or ulcers caused by other agents; recent evidence suggests that early colonization of small wounds by H. ducreyi is the more likely mechanism (29). The best clinical description of H. ducreyi-associated CU is in reference (30). Most H. ducreyi-associated CU occurs on the lower extremity. Compared to CU associated with dual H. ducreyi and Treponema pallidum subsp. pertenue infections and T. p. pertenue alone, H. ducreyi ulcers tend to be more tender, are significantly more likely to have shorter diameters, and are significantly less likely to be round, deep, uniform in color with a granulating ulcer bed, or have indurated edges (Fig. 2) (30). These features may aid in clinical diagnosis, but there is so much overlap among these three entities that molecular testing is required for a definitive diagnosis.

Fig 2.

Leg ulcers in children in Papua New Guinea. (A) Irregular shallow ulcer that was positive for H. ducreyi DNA. (B) Round, indurated, deep ulcer that was positive for T. p. pertenue DNA; note flies on the ulcer. (Both images courtesy of Camila Gonzalez-Beiras and Oriol Mitja; reproduced with permission.)

DIAGNOSTIC TESTING FOR GU AND CU

As reviewed in reference (31), culture and nucleic acid amplification tests are the cornerstones of diagnostic testing for H. ducreyi infections. The best characterized nucleic acid amplification test is the Roche Multiplex PCR (M-PCR) assay, which has a resolved sensitivity of 95–98% and a specificity of 99.6% for H. ducreyi in chancroid (32, 33). Compared to M-PCR, the resolved sensitivity for the most sensitive (dual plate) culture system in chancroid is approximately 75% (32, 33). Compared to PCR-based tests, the clinical diagnosis of chancroid is neither sensitive (range, 52-75%) nor specific (range, 51.9-74.5%) (2); definitive diagnosis requires either a positive culture or PCR test. Due to a lack of laboratory infrastructure, cultures have seldom been used in the diagnosis of H. ducreyi-associated CU (9–11); most studies have used PCR-based tests for detection, using either primers developed for the M-PCR assay or investigator-designed reagents.

Given the declining prevalence of chancroid in most but not all formerly endemic areas, routine screening of patients with GU for H. ducreyi is not currently cost effective. A reasonable approach is to use culture or PCR-based tests to screen GU patients when syndromic management for syphilis and herpes fails and chancroid is suspected and to modify syndromic management if chancroid is found (34). For CU, PCR testing, when available, is considered diagnostic, although the sensitivity and specificity of these tests in this setting are unknown.

ANTIBIOTIC RESISTANCE AND THERAPY

Given the decline of chancroid and the utilization of PCR-based tests as opposed to culture for both GU and CU, there are little current data on antibiotic resistance in H. ducreyi. In the 1990s, plasmid-mediated resistance to ampicillin, chloramphenicol, tetracyclines, and sulfonamides were reported in most GU isolates (28, 35, 36). These plasmids are homologous to R factors present in Gram-negative enteric organisms and Neisseria and Haemophilus species strains (28). Some GU strains contained an integrated conjugative resistance element (ICE) that belongs to a family of conserved genomic islands that have a common evolutionary origin with strains of numerous different bacterial species, including H. influenzae (37). Antibiotic, metal, and antiseptic resistance genes, along with other accessory genes, are typically found in ICE (37). The 1990s saw a rise in trimethoprim resistance, whose mechanism is unknown (28, 35). Macrolide, quinolone, and third-generation cephalosporin resistance was not reported, and drugs belonging to these classes are still recommended for treatment of chancroid (28, 35, 38) (Table 1). Of 17 contemporary CU strains sequenced, four contained tetB, two contained catS, and only one contained a bla determinant (18); the MIC for penicillin done on three bla-negative strains was 0.16 µg/mL (9, 18). One amoxicillin-resistant CU strain isolated from a lesion acquired in Indonesia has also been reported (39).

TABLE 1.

Treatment of chancroid from different societies (recommended, not recommended, and alternative)^a

Medication	Dose	US Public Health Service	World Health Organization	UK Clinical Effectiveness Group
Azithromycin	1 g p.o. in a single dose	Recommended	Recommended	Recommended
Ciprofloxacin	500 mg p.o. bid for 3 days	Recommended	Not Recommended (Alternative: Norfloxacin)	Recommended
Ceftriaxone	250 mg i.m. in a single dose	Recommended	Not Recommended (Alternative: Norfloxacin)	Recommended
Erythromycin base	500 mg p.o. q.i.d. for 7 days	Not Recommended	Not Recommended	Recommended
Norfloxacin	800 mg p.o. as a single dose	Not Recommended (Alternative: Ceftriaxone)	Recommended	Not Recommended
Spectinomycin	2 g i.m. in a single dose	Recommended	Not Recommended	Not Recommended

^a The recommendations are indicated as “Recommended” or “Not Recommended,” and for the “Not Recommended” medications, alternative recommendations are mentioned in parentheses.

There have been no antibiotic treatment trials for chancroid since the 1990s. On Lihir Island, of 131 children with H. ducreyi-associated CU who were treated with a single dose (30 mg/kg) of azithromycin, 94.7% of ulcers either healed or improved 2 weeks after treatment (40). In a randomized trial conducted in Ghana and the Karkar District of Papua New Guinea (PNG), the clinical cure rate at 4 weeks was 100% for those treated with 20 mg/kg (N = 51) or 30 mg/kg (N = 51) of oral azithromycin (41), confirming that azithromycin is an effective treatment for H. ducreyi-associated CU. No other regimens have been tested for CU.

MOLECULAR EPIDEMIOLOGY OF GU AND CU

Given that M-PCR is the most sensitive test for the diagnosis of chancroid and has high specificity (32, 33), we reviewed 42 studies that used M-PCR or other PCR-based tests on persons with GU attending STD clinics from 1992 to 2021 (Table 2). We grouped the studies by regions, specifically Africa, Asia, South America and the Caribbean, and North America, Europe, and Australia (Table 2). In some studies, either men or women were excluded; there were a total 9,780 patients tested, including 7,431 (76%) men and 2,349 (24%) women. Of the patients sampled, 1,149 (11.8%) had lesions that were positive for H. ducreyi DNA; of the 1,149 samples, 213 (18.5%) also contained herpes simplex and/or Treponema pallidum subsp. pallidum DNA. Thus, co-infections among H. ducreyi and other agents that cause GU are common.

TABLE 2.

H. ducreyi prevalence in surveys of patients with genital ulcers attending STI clinics as determined by PCR^a

Region	Year	No. tested		% positive for H. ducreyib	Reference
		M	F
Africa
Dakar, Senegalc	1992	41	5	56	(42)
Lesotho, South Africa	1993	69	36	56	(33)
Durban, Johannesburg,   Cape Town, South Africa	1993–1994	538	E	32	(43)
Carletonville, South Africa	1993–1994	232	E	69	(44)
Antananarivo, Madagascar	1997	139	57	33	(45)
Carletonville, South Africa	1998	186	E	51	(44)
Lilongwe, Malawi	1998–1999	94	43	30	(34)
Dar es Salaam & Mbeya,   Tanzaniac	1999	54	48	21	(46)
Dar es Salaam, Tanzaniac	1999–2001	210	91	4	(47)
Durban, South Africa	2000–2001	438	149	10	(48)
Three cities, Botswana	2001–2002	79	58	0.7	(49)
Central African Republic /   Ghana	2003–2005	E	422	0.7	(50)
Durban, South Africa	2004	162	E	1.8	(51)
Lilongwe, Malawid	2004–2006	294	104	15.1	(52)
Rakai District, Ugandad	2002–2006	50	50	2	(53)
Gauteng Province, South   Africac	2005–2006	615	E	1.6	(54)
Bissau, Guinea-Bissauc	2006–2008	E	155	12.9	(55)
Kampala, Ugandad	2008–2009	E	62e	6.0	(56)
Lusaka, Zambiad	2010	100	100	0	(57)
Maputo, Mozambique	unknown	48	28	4.0	(58)
Johannesburg, South   Africad	2007–2015	681	90	0.5	(59)
Zimbabwe	2014–2015	100	100	0	(60)
Eastern Cape, South Africad	2018–2019	27	78	8.6	(61)
Lilongwe, Malawid	2021	32	18	18	(62)
Asia
Pune, India	1994	277	25	28	(63)
Chiang Mai City, Thailand	1995–1996	8	30e	0	(64)
Shanghai & Chengdu, PRC	2000	204	23	0	(65)
Karnataka State, India	2004–2006	206	66	0	(66)
India	2008–2009	192	E	0.5	(67)
India (seven states)	unknown	219	18	0.4	(68)
South America, Caribbean
Lima, Peru	1994–1995	63	E	5	(69)
Santo Domingo,   Dominican Republic	1995–1996	81	E	26	(69)
Kingston, Jamaica	1996	252	52	24	(70)
Manaus, Brazil	2008–2009	368	66	0	(71)
Havana, Cuba	2012–2015	113	E	0	(72)
Brazil (nationwide)d	2018–2020	206	E	0	(73)
North America, Europe, Australia
New Orleans, Louisiana	1992–1994	298	E	22	(32)
Jackson, Mississippi	1994–1995	111	32	39	(74)
Ten cities, United States	1996	351	165	3	(75)
Amsterdam, The   Netherlandsc	1996	221	151	1	(76)
Queensland, Australia	2002	38	26	0	(77)
Graz, Austria; Zurich,   Switzerlandd	2012–2013	34	1	0	(78)
Total		7,431	2,349	11.7

^a All studies used M-PCR (32) except where indicated. M, males; F, females; E, excluded.

^b Includes specimens positive for H. ducreyi DNA and for both H. ducreyi and T. pallidum subsp. pallidum or herpes simplex virus DNAs.

^c Studies used investigator designed reagents.

^d Studies used real-time PCR.

^e Sex workers.

In Africa, from 1992 until 1999, chancroid was highly prevalent in patients with GU, ranging from 21% to 69% (Table 2). After 2001, the prevalence of chancroid was much lower in most formerly highly endemic areas, generally ranging from 0% to 6% (Table 2). In areas that had been sampled prior to and after 2001, such as Johannesburg, the prevalence of chancroid fell from 32% to 0.5%. Although the prevalence of chancroid in Lilongwe Malawi fell from 30% in 1999 to 15.1% in 2006, chancroid has persisted in this area with a prevalence 18% in 2021. Other regions where chancroid prevalence still exceeded 6% after 2001 include Guinea-Bissau and the Eastern Cape (Table 2).

In Asia, chancroid prevalence was as high as 28% in Pune, India, in 1994, but was rare (0–0.5%) in other areas of India surveyed after 2001 (Table 2). Conversely, locations in Thailand and the People's Republic of China had negligible or zero prevalence rates throughout this period. In South America and the Caribbean, chancroid prevalence ranged from 5% to 26% in three surveys done in the 1990s but was not detected in three surveys done after 2001 (Table 2). In North America, there were urban outbreaks tied to sex work in the early 1990s, but there were no further outbreaks by 1996. More recent studies in European cities showed no detectable H. ducreyi. Similarly, studies in Australia and the Netherlands reported very low prevalence rates (0–1%).

In summary, H. ducreyi was a highly prevalent cause of GU in certain regions of Africa, Asia, and the Caribbean in the 1990s and was absent in North America and Europe except for sporadic outbreaks. Due to syndromic management and the lack of diagnostic testing, the epidemiology of chancroid is currently undefined, but most surveys have shown a dramatic decline in H. ducreyi as a cause of GU in formerly endemic areas. As will be discussed in the section on transmission dynamics, the persistence of chancroid in some regions implies that there is a reservoir of infected sex workers in that community.

In published surveys for active yaws done between 2001 and 2013 in equatorial regions of the South Pacific islands and Africa, 0.5–14.5% of children had painful exudative cutaneous ulcers (CU) that were attributed to T. p. pertenue (79, 80). Once single dose oral azithromycin was found to be as effective as benzathine penicillin for the treatment of yaws (81, 82), clinical trials were done to examine whether mass drug administration (MDA) of oral azithromycin and subsequent case finding and treatment every 6 months could eliminate yaws in endemic areas. However, molecular diagnostic tests done on swabs of ulcers obtained prior to MDA show that the etiology of CU is multifactorial (4, 30, 83–89) (Table 3). For example, on Lihir Island in Papua New Guinea (PNG), 47% of ulcers were positive for H. ducreyi, 21% for T. p. pertenue, and 13% for both H. ducreyi and T. p. pertenue DNAs. However, ~20% were T. p. pertenue and H. ducreyi DNA negative (30); the latter group was defined as having idiopathic ulcers (IU). By microbiome analyses, subsequent studies on Lihir Island show that Streptococcus pyogenes is the most abundant DNA in IU and is present in ~39% of the IU samples (85, 90). Even within a single country, there is regional variation in organisms associated with CU. In the Oti region of Ghana, 40% of ulcers are positive for Leishmania species, 67% for T. p. pertenue, and 73% for H. ducreyi DNAs; co-infections with more than one organism occur in 68% of cases, and IU comprised only 8% of cases. In contrast, H. ducreyi is detected in 9% of CU samples in the Ashanti District of Ghana, while HSV-1 and treponemal DNAs are detected in only 1.8% and 0.9% of the samples, respectively (91). In the Yendi and Savelugu-Nanton Districts of Ghana, which had received multiple rounds of annual MDA of azithromycin for trachoma 7 years before the survey was conducted, H. ducreyi is detected in 8% of CU samples, but no T. p. pertenue is detected. As none of the children with CU had positive serology for treponemal infection, the data suggest that yaws is absent in these districts and there is a high rate of IU; however, tests for Leishmania species were not included in this study (92) or in the Ashanti District survey (91). For studies that reported demographic data for children with H. ducreyi-associated CU, the median age ranges from 9 to 10 (30, 86, 88, 89, 92) and 59.5–62% are males (30, 84, 86, 88); however, two studies reported similar infection rates in both sexes (21, 40) . Taken together, the data show that CU and GU are similar in that they are caused by multiple (albeit different) organisms, mixed infections are common, and there are regional variations in the etiologies of both syndromes.

TABLE 3.

H. ducreyi prevalence in population surveys of children with cutaneous ulcers as determined by PCR

Region	Year	No. tested	% positive for H. ducreyia	References
Prior to community-based MDAb of azithromycin
Lihir Island, PNGc	2013–2014	90	60	(30, 83)
Western and Choiseul Provinces, Solomon Islands	2013	41	31.7	(84)
Ghana	2013	179	27.3	(4)
Vanuatu	2013	176	38.6	(4)
Lolodorf and Lornie Health Districts, Cameroon	2017–2019	111	49.6	(86)
New Ireland Province, PNG	2018–2019	471	25.3	(87)
Oti Region, Ghana	2019	101	73.3	(88)
14 Health Districts, Cameroon	2021–2022	271	30.3	(89)
Ashanti Region, Ghana	2021	110	9.0	(91)
After community-based MDA of azithromycin
Yendi and Savelugu-Nanton Districts, Ghanad	2014	90	8.8	(92)
Lihir Island, PNGe	2013–2015	198	60.1	(83)
Lihir Island, PNGf	2016–2018	275	45	(85)
New Ireland Province, PNGg	2018–2019	361	28	(87)
New Ireland Province, PNGh	2018–2019	116	28.4	(87)

^a Includes specimens positive for H. ducreyi DNA and for both T. pallidum subsp. pertenue and H. ducreyi DNAs.

^b MDA, mass drug administration.

^c PNG, Papua New Guinea.

^d 7 years after four to six annual rounds of MDA for trachoma.

^e 6 and 12 months after one round of MDA for yaws.

^f 36, 42, and 48 months after one round of MDA for yaws.

^g 6, 12, and 18 months after one round of MDA for yaws.

^h 6 months after three rounds of MDA for yaws.

In cohorts followed before and after MDA of azithromycin on Lihir Island and New Ireland Province in PNG, the proportion of children with CU was reduced by ~90%, but the percentage of H. ducreyi -associated CU was unchanged (83, 85, 87) (Table 3). Although azithromycin is an effective treatment for H. ducreyi -associated CU (40), azithromycin is not effective in eradicating H. ducreyi on a community level. In Cameroon and PNG, 8.6–21% of asymptomatic children tested have detectable H. ducreyi DNA in swabs of their lower limbs (89, 93); H. ducreyi was isolated from two children in one study, proving asymptomatic colonization (93). Flies are found on CU lesions (Fig. 2) and bed sharing is common in endemic areas; 90% of flies and 33% of bedsheets sampled have detectable H. ducreyi DNA (93). When exposed to suspensions of H. ducreyi, the common housefly can carry viable bacteria on its appendages for up to 90 minutes, suggesting that flies may be capable of mechanical transmission of the organism from person to person (94). Asymptomatic colonization and environmental factors likely explain why H. ducreyi-associated CU escapes antibiotic pressure and persists in these regions (89, 93).

Why were organisms other than T. p. pertenue not recognized as a cause of CU previously? Clinical microbiology laboratories are generally unavailable in yaws-endemic regions, and empiric benzathine penicillin has been the standard treatment for suspected yaws since the late 1940s. As stated earlier, most CU strains sequenced do not contain a bla determinant (9). To date, S. pyogenes has not been reported to be penicillin resistant; it is likely that empiric therapy with penicillin for yaws cured most CU, allowing H. ducreyi and S. pyogenes to go unrecognized.

A map of countries in which H. ducreyi-associated CU has been documented and in which yaws is classified as endemic in 2024 by the World Health Organization (https://www.who.int/data/gho/data/indicators/indicator-details/GHO/status-of-yaws-endemicity) is provided in Fig. 3. Three countries (Fiji, Samoa, and Sudan) with case reports of H. ducreyi-associated CU are currently classified as “previously yaws endemic, current status unknown.” Six countries in which surveys (Cameroon, Ghana, PNG, Solomon Islands, and Vanuatu) or a case report (Indonesia) documented the presence of H. ducreyi-associated CU are classified as yaws endemic. Ten yaws-endemic countries (Benin, Cote d' Ivore, Central African Republic, Congo, Democratic Republic of Congo, Liberia, Maylasia, Philippines, Timor Leste, and Togo) have not been investigated for the presence of H. ducreyi-associated CU. This lack of data represents a major gap in our understanding of the global epidemiology of H. ducreyi-associated CU.

Fig 3.

Map of countries with documented H. ducreyi-associated CU and countries classified by the World Health Organization as yaws endemic in 2024. Red, countries with case reports of H. ducreyi-associated CU that are currently classified as “previously endemic, current status unknown” for yaws. Yellow, countries classified as yaws endemic in which H. ducreyi-associated CU has been documented. Blue, countries that are yaws endemic but have yet to be investigated for H. ducreyi-associated CU.

PATHOGENESIS OF H. DUCREYI IN HUMAN VOLUNTEERS AND ITS RELATIONSHIP TO GU AND CU

Most of what is known about H. ducreyi pathogenesis is derived from experiments in which the Class I GU strain 35000HP (HP, human passaged) and its derivatives are inoculated into the skin overlying the deltoid muscle of healthy adult human volunteers [reviewed in references (95–97)]. Neither Class II GU strains nor CU strains have been evaluated in this model. Placement of 10⁶ H. ducreyi CFU on intact skin does not cause infection (98); puncture wounds made by the tines of an allergy testing device are required to initiate infection, and the estimated infectious dose is between 1 and 100 CFU (99, 100). These findings are consistent with the ideas that chancroid is initiated by microabrasions that occur during sex with an infected partner and that CU results from contamination of traumatic wounds by H. ducreyi.

In the model, papules develop within 24 h of inoculation and either spontaneously resolve or evolve into pustules 2–5 days later (Fig. 4). Pustules enlarge and generally become mildly pruritic and painful 7–14 days later. Within 24 h of inoculation, H. ducreyi co-localizes with collagen and fibrin that are deposited in the wounds; neutrophils and macrophages surround but fail to ingest the bacteria (101, 102). An abscess forms that eventually erodes through the epidermis. Below the abscess is a collar of macrophages and regulatory T cells and a dermal infiltrate of macrophages, CD4 and CD8 T cells, natural killer cells, and dendritic cells (DC). This histopathology and the relationships between H. ducreyi and host cells or components are maintained in chancroidal ulcers (103), but there have been no reports to date on the histopathology of CU. Thus, the human infection model simulates the first 2 weeks of natural chancroid but does not provide information about the ulcerative stage of disease or lymphadenitis.

Fig 4.

Host effect on the outcome of experimental infection. (A) Papules and pustules (B) that formed 1 day and 7 days after inoculation of three sites with live 35000HP in volunteer 485. (C) Papules that formed 1 day after inoculation of three sites with live 35000HP and resolved (D) by day 7 in volunteer 497.

In the human model, the most significant predictors of outcome are host and sex. The outcomes (pustule vs. resolved) of multiple infected sites within a subject are not independent, suggesting an overall host effect on the sites (Fig. 4) (104). The host effect was confirmed in reinfection trials, which showed that some volunteers repeatedly form pustules while others repeatedly resolve infection, presumably by their ability to overcome the antiphagocytic properties of the organism (104, 105). The mechanism for the host effect may be due to differential dendritic cell responses to the organism (97, 106). When infected with H. ducreyi, monocyte-derived DC from resolvers promote a Th1 response that may overcome the antiphagocytic properties of the organism and facilitate bacterial clearance, while DC from pustule formers promote a combined Th1 and regulatory T cell response that likely contributes to phagocytic failure (106). As only persons with ulcers generally seek medical attention, whether GU or CU spontaneously resolve in naturally infected persons is unknown.

In addition to the host effect, data obtained from infection of 354 volunteers who were followed until they formed pustules or resolved infection show that inoculated sites in men and women become infected (form papules) at equal rates, but pustules form at male sites 1.7 (95% CI: 1.3–2.4) fold that of sites in women (96, 100) (unpublished observations), consistent with the excess male to female ratio in chancroid (2-3:1) and in the majority of studies of CU (1.5:1). As will be discussed below, other factors such as number of sex partners and prevalence of trauma likely contribute to sex differences in GU and CU, respectively.

TRANSMISSION DYNAMICS AND THE POTENTIAL FOR ERADICATION OF CHANCROID

The transmission dynamics of STIs are calculated by the equation (R₀= βcD) where R₀ is the reproductive rate, β is the transmission rate per sex act, c is the sex partner change rate per year, and D is the average duration of infectiousness in years (107). For an STI to remain endemic in a population, R₀ must be >1. For H. ducreyi, D is estimated to be ~4 weeks or 0.08 years based on the observations that viable H. ducreyi are recovered from swabs of experimental papules and pustules (108) and that medical attention is sought for chancroid after 1–3 weeks of ulcerative symptoms (7, 22). For H. ducreyi, β is unknown; in one small study, 70% of women who were secondary contacts of men with chancroid had genital ulcers (109), suggesting that the upper limit for β is 0.7. For β = 0.7, the calculated c is 18; for β = 0.35, c = 36, and for β = 0.18, c = 72 (110). These estimates suggest that chancroid can be maintained in a community only by the presence of infection in highly sexually active populations such as sex workers, consistent with epidemiological data suggesting that sex workers are the major clinical reservoir for chancroid (24, 109).

Given the transmission dynamics of H. ducreyi, eradication of chancroid from sex workers should result in disappearance of the disease from the community (1). In Thailand, policies for 100% condom use and presumptive treatment of sex workers with quinolones led to a 95% decrease (from 30,000 cases to less than 2,000 cases) in chancroid in the 1990s (111). In the human challenge model, a single dose (1 g) of oral azithromycin prevents volunteers from developing disease after weekly inoculations for ~2 months, due to recirculation of azithromycin in intracellular compartments, which results in an elimination half-life of 7.5 days (112). In a South African mining community, monthly presumptive antibiotic treatment of sex workers with a 1 g dose of oral azithromycin for 3 months, condom promotion, and education directed toward prevention resulted in 85% decline in GU in the sex workers and a 78% decline in GU in the general population over a 9 month follow-up period (1, 113). Taken together, these studies suggested that chancroid is an eradicable disease, either by promotion of targeted antibiotic treatment of sex workers or by widespread application of syndromic management, both of which likely contributed to the dramatic decline in chancroid in formerly endemic areas (Table 2).

ASYMPTOMATIC COLONIZATION, ENVIRONMENTAL FACTORS, AND THE FAILURE TO ERADICATE CU

Persons who live in yaws-endemic regions are said “to live on the ground,” meaning that there is little sanitation, poor hygienic conditions, exposure to livestock such as swine and chickens who share their environment, and a propensity to have traumatic wounds. Interestingly, H. ducreyi DNA is detected by metagenomic sequencing of swine and poultry manure processed for compost in South China (114); whether H. ducreyi DNA or viable organisms are present in livestock manure in CU-endemic areas remains to be determined. To date, H. ducreyi DNA has been detected on intact skin of the legs in up to 20% of asymptomatic children residing in both CU and non-CU households, flies, and bed linens (89, 93); other environmental sources (soil, animals, and water sources) have yet to be evaluated. In contrast, only 2% of sex workers without GU on speculum examination have detectable H. ducreyi DNA on cervical swabs (24), suggesting that asymptomatic colonization is relatively rare in the female genital compartment vs. the skin in GU- and CU-endemic areas, respectively. This could be due to differences in immune activity or microbiomes of the vaginal compartment vs. the skin surface, especially under the suboptimal hygienic conditions as described above.

The MICs of azithromycin for H. ducreyi are in the range of 0.0005 to 0.004 µg/mL (112). A single 1 g dose of oral azithromycin achieves median concentrations in cervical mucus of 2.67 µg/mL, 1.26 µg/mL, and 0.15 µg/mL 1 day, 7 days, and 14 days, respectively, after administration, which is sufficient to clear H. ducreyi infection and colonization of the urogenital compartment (115). As azithromycin is concentrated primarily in fibroblasts (112), one would not expect azithromycin to clear H. ducreyi colonization of intact skin. Application of a single locus typing system to CU swabs containing H. ducreyi DNA obtained prior to and 1 year and 2 years after MDA of azithromycin on Lihir Island showed that the composition and geospatial distribution of H. ducreyi strain types changed little over time and no evidence of selection pressure, indicating that antimicrobial treatment is not sufficient to eliminate colonization and control CU (20). In contrast, T. p. pertenue is thought to be transmitted primarily by skin to skin contact with an infected individual, and MDA of azithromycin did reduce T. p. pertenue strain diversity on Lihir Island (20, 116).

TOPICAL TREATMENTS AND VACCINES FOR THE PREVENTION OF CU

Although chancroid may be eradicated by antimicrobials, the persistence of H. ducreyi-associated CU despite antimicrobial therapy has spurred interest in whether topical treatments or vaccines might prevent CU. Resveratrol, which is produced by flowering plants, is bactericidal for H. ducreyi but has not been tested clinically (117). Ficus septica exudate is a traditional medicine used in PNG for treatment of wounds; the alkaloid ficuseptine is the principle antibacterial compound contained in the exudate (29). In a randomized cluster trial of 150 school children with skin ulcers <1 cm in diameter, in which 50 children in each of three schools received two topical treatments given over two consecutive days of either soap and water, a 0.01% chlorhexidine antiseptic cream, or unprocessed ficus exudate, the exudate was noninferior to the other two treatments with ~75% participants having improved or healed ulcers at 14 days of follow-up (29). Prior to treatment, few (0–6%) of the ulcers contained T. p. pertenue DNA but most (92–100%) contained H. ducreyi DNA, suggesting that H. ducreyi CU likely originates in small traumatic wounds (29).

In the human challenge model, mutant vs. parent trials have been used to identify bacterial components that are required for infection and could serve as vaccine candidates, particularly if they raised protective antibodies against this extracellular pathogen [reviewed in references (95–97)]. In these double blind trials, volunteers are inoculated at multiple sites on one arm with the mutant and on the other arm with the parent and serve as their own controls for host and sex effects. Mutants are categorized as virulent (form pustules at doses equivalent to the parent), partially attenuated (form pustules at doses twofold or threefold that of the parent, but not at doses equivalent to the parent), or as fully attenuated (unable to form pustules even at doses 10-fold that of the parent). Of 37 mutants tested, 10 are fully attenuated (97, 118–120) (unpublished); two secreted proteins (LspA1 and LspA2) that inhibit phagocytosis; four outer membrane proteins that promote hemoglobin uptake (HgbA), adherence to collagen (NcaA), outer membrane stability (PAL), and resistance to complement mediated killing (DsrA); one antimicrobial peptide transporter system (SapABC); and one family of fimbriae (Flp1,2,3) that confer the ability of the bacterium to form aggregates are absolutely required for pustule formation (97). BLAST analysis of CU and GU genomes show that all strains contain the genes absolutely required for virulence and that, except for some polymorphisms in dsrA, hgbA, and ncaA, these genes are highly conserved (17, 18, 121) and could serve as vaccine candidates.

Preclinical studies employing Class I and Class II GU strains in a swine model of chancroid indicate that a recombinant form of the N-terminal passenger domain of DsrA protects against infection with a homologous Class I strain (122), while purified native HgbA protects against infection with a homologous Class I strain but not a Class II strain (123, 124). Passive transfer of polyclonal antiserum from pigs immunized with HgbA also protected against homologous Class I but not heterologous Class II strain challenges, suggesting that the mechanism of protection is antibody-mediated (125). Antibodies elicited by immunization with HgbA bind to the surface of H. ducreyi and partially block hemoglobin binding, suggesting that the protective mechanism is through the restriction of heme/iron acquisition (125). The data also suggest that a bivalent HgbA vaccine will be necessary to target both classes. Although likely, whether the preclinical data indicating protection from infection by GU strains can be extrapolated to protection from infection caused by CU strains remains to be determined. In addition, the dose of H. ducreyi required to cause disease in the swine model is 10⁴ CFU, which is several orders of magnitude higher than the dose (1–10² CFU) required to initiate infection in humans (95). In swine, the number of bacteria recovered from lesions decreases to 10³ CFU 48 h after inoculation, but viable bacteria persist in swine lesions for up to 17 days; in humans, H. ducreyi replicates to ~10⁵ CFU in pustules (126–128). Thus, whether vaccine candidates that show promise in the swine model will prove to be effective in humans also remains to be determined.

Recently, a bioinformatics approach (reverse vaccinology) was utilized to identify potential H. ducreyi vaccine targets (129). Using the genome sequences of 28 H. ducreyi strains to identify a core genome for the species and algorithms to identify conserved surface exposed targets with suitable MHC I and MHC II binding properties, 13 putative vaccine candidates were identified, including HgbA (129). Of the remaining 12 candidates, a formate efflux transporter, encoded by focA, was subsequently found to be dispensable for infection in the human challenge model and therefore may not be a viable candidate (119). The other candidates have yet to be characterized either for their roles in virulence or for vaccine efficacy in preclinical models.

CONCLUSIONS

GU and CU strains of H. ducreyi share a core genome and many aspects of biology but have evolved in distinct ecological niches. Given its requirement for person-to-person transmission and the presence of infection in highly sexually active populations, chancroid may be eradicated by syndromic management and targeted antimicrobial prophylaxis of sex workers. Although vaccines may not be necessary for eradication of chancroid, they may be necessary for eradication of H. ducreyi-associated CU due to asymptomatic skin colonization and environmental factors. Topical treatments of small wounds with antiseptics or native plant extracts may help prevent infection. Efforts should be made to evaluate other yaws-endemic countries for the presence of H. ducreyi-associated CU, the environment for additional reservoirs of CU strains, and to eliminate these reservoirs and asymptomatic colonization by improvements in hygiene and sanitation; such public health measures could serve to eradicate this painful, disfiguring infection in children.

Biographies

Jaffar A. Al-Tawfiq received an M.B.B.S. from King Faisal University, Dammam, Saudi Arabia and completed a residency in Internal Medicine and an Infectious Diseases fellowship at Indiana University School of Medicine, Indianapolis, IN, USA. He joined the faculty of the Saudi Aramco Medical Services Organization (SAMSO), Saudi Arabia, as a consultant of Internal Medicine and Infectious Diseases and later served as the head of general internal medicine and the head of infection control. He is currently, the director of accreditation and infection control at Johns Hopkins Aramco Healthcare, Saudi Arabia. He is also an adjunct professor of medicine and infectious diseases at Indiana University. During his fellowship, his interest was the pathogenesis of H. ducreyi, and he worked with Dr. Spinola focusing on basic and clinical studies of H. ducreyi in human volunteers. His current research interests include the epidemiology of emerging infectious diseases, healthcare associated infections and global health.

Stanley M. Spinola received a B.A. from Brown University, a M.D. from Georgetown University and completed a residency in Internal Medicine and Pediatrics and a fellowship in Infectious Diseases at the University of North Carolina. He joined the faculty of SUNY-Buffalo in 1987 and came to Indiana University in 1993 to join the Division of Infectious Diseases. He served as ID Division Director (1995-2010), Chair of the Department of Microbiology and Immunology (2010-2019) and is a Professor of Microbiology and Immunology, Medicine and Pathology and Laboratory Medicine. His research on bacterial pathogenesis began in 1984 and on H. ducreyi in 1987. In 1992, he developed a model in which human volunteers are infected on the skin of the upper arm with H. ducreyi. His research focuses on the interplay between H. ducreyi and the human host, the ethics of human infection models, and the causes of cutaneous ulcers in children.