Clinical and Histopathological Presentation of Buruli Ulcer in Experimentally Infected Grasscutters (Thryonomys swinderianus)
P Addo, B Adu-Addai, M Quartey, M Abbas, I Okang, E Owusu, D Ofori-Adjei, B Awumbila
animal model, buruli ulcer infection, clinical features, grasscutter, histopathological features
P Addo, B Adu-Addai, M Quartey, M Abbas, I Okang, E Owusu, D Ofori-Adjei, B Awumbila. Clinical and Histopathological Presentation of Buruli Ulcer in Experimentally Infected Grasscutters (Thryonomys swinderianus). The Internet Journal of Tropical Medicine. 2006 Volume 3 Number 2.
Buruli ulcer (BU) is a skin disease caused by Mycobacterium ulcerans, a toxin-producing mycobacterium. BU manifests as papule, plaque, nodule, oedema and undermined ulcer with accompanying complications such as contracture deformities and osteomyelitis. Its mode of transmission and pathogenesis are unclear and effective treatment is not available. Therefore animal modelling, among other strategies is being undertaken to elucidate the problems. This study investigated the grasscutter (Thryonomys swinderianus), a hystricomorph rodent as a BU animal model. Grasscutters were inoculated subcutaneously with Mycobacterium ulcerans and developed progressive skin lesions: erythema, papule, nodule, oedema, undermined ulcer and contracture. The lesions were accompanied by histopathological changes, specifically: coagulative necrosis, mixed inflammatory reactions, myolisis, perineuritis, neurogenic muscular atrophy and osteomyelitis; some of which were accompanied by extracellular and intravascular acid fast bacilli (AFBs). Our findings show that grasscutters are susceptible to Mycobacterium ulcerans; they develop clinical, microbiological and histopathological lesions that mimic BU in humans and therefore are potential BU animal models.
Department of Animal Experimentation, Noguchi Memorial Institute for Medical Research
Source of Funding
Noguchi Memorial Institute for Medical Research
Buruli ulcer (BU) is a poorly understood disease caused by
In the face of the enigma that BU presents, the World Health Organization (WHO) has recommended that animal studies should be undertaken worldwide to elucidate the pathogenesis and mode of transmission of the disease and to provide valuable new approaches for its diagnosis, treatment, management and general control (WHO, 2001c). Though mice, rats, guinea pigs and the nine-banded armadillo are BU animal models (Walsh et al, 1999), each of them has limitations in replicating the whole spectrum of features presented in humans, thus leaving a lot of gaps in knowledge on the pathogenesis, transmission, diagnosis, treatment and general control of BU (Pattyn & Royacker, 1965; Reed et al, 1974; Krieg et al, 1974). Despite the limitations of the available animal models, the mouse, (which is the most frequently used model) has provided a lot of useful information on both the pathogenesis (Pattyn & Royacker, 1965; Reed et al, 1974; Addo et al, 2005; Goto et al, 2005) and treatment of BU (Bentoucha et al, 2001; Dega et al, 2000 & 2002). This study is a follow-up to a BU mouse study undertaken at the Noguchi Memorial Institute for Medical Research (NMIMR). In the said study, though the mouse presented a plethora of BU lesions (Addo et al, 2005), confirming its usefulness as a BU animal model, it did not manifest undermined ulcer which is the hallmark of BU. The mouse was also too small to facilitate proper clinical evaluation of the sores that developed, which also unfortunately progressed rapidly to necrosis, resulting in the destruction of the limb. These findings gave support to the assertion that better BU animal models need to be found (Walsh et al, 1999; Addo et al, 2005). This paper reports on an attempt to identify another animal that would present in addition to other BU lesions, undermined ulcer, the hallmark of BU.
The grasscutter (
Materials and Methods
Twenty-six, 4-month old grasscutters produced in the conventional animal facility of the Department of Animal Experimentation were used for the study. They had been born and bred in an animal facility in which all conventional animal colonies are microbiologically monitored quarterly, specific pathogen free colonies are monitored monthly, and the animal facilities (conventional, barrier) are environmentally monitored monthly. A month before the inception of the study the microbiological status of the selected animals was ascertained by use of non-invasive monitoring methods namely; swabbing of eyes, nostrils and ears; sampling of faeces, blood and urine for bacteria, mycoplasma, fungi and ecto and endoparasites. Thereafter they were transferred to the animal experimentation facility (level 2 containment) to acclimatize. The animals were separated into two experimental groups and one control group. Each of the experimental groups consisted of 5 males and 5 females while the control group consisted of 3 males and 3 females. The animals were maintained and manipulated in accordance with the Institutional animal care and use guidelines.
Research Clearance and Biosafety Considerations
Institutional research clearance was obtained before the study was undertaken. The study was carried out in a level 2 biosafety laboratory. All microbiological manipulations were conducted in a class IIA/B3 biosafety cabinet and the technical team observed all institutional biosafety guidelines for protection of personnel and laboratory.
Isolation of from Tissue
The tissue was decontaminated and inoculated onto Lowenstein-Jensen (L-J) medium within 24 hours of its collection from the hospital. Briefly, the tissue was divided into two portions; one portion was homogenized for primary culture of
Preparation of Inoculum.
The primary isolate was subcultured on L-J slants at 32 C for eight weeks, during which period they were examined for growth and contamination. The subcultured
Experimental Infection of Grasscutters
The two experimental groups were inoculated subcutaneously with
Management of Grasscutters after Experimental Infection
The experimental and control groups of grasscutters were managed in the same manner. In brief, the three groups of animals were kept in the same containment facility (under negative pressure) but in three separate negative pressure air racks. The animals were provided with HEPA-filtered air, maintained at 24-25ºC ambient temperature, 55-65% relative humidity and a 12-12 hour light-dark cycle (by means of an automatic lighting system). Each grasscutter was kept in a metal cage (H: 50cm, W: 40cm, L: 40cm) provided with a floor mesh, feeder and drinking bottle. All three groups of grasscutters were fed daily with autoclaved commercial rodent feed pellets (Ghana Agro Food Complex), sanitized Guinea grass (
Evaluation of Grasscutters after Inoculation
Clinical and Microbiological Evaluation
The experimental and control groups of grasscutters were observed for post-inoculation changes (erythema, papule, plaque, oedema and ulcer) at the points of inoculation in particular and shaved thighs in general. The animals were initially observed daily for one month and thereafter thrice weekly (Mondays, Wednesdays and Fridays). The faeces, urine and the blood of the animals in Groups 1 and 2 were examined monthly for AFBs after ZN staining. Whenever any discharge (exudate, pus) or scab was available they were ZN-stained for the detection of AFBs. The discharge was also cultured on different types of media (blood agar, MacConkey agar, Mannitol salt agar,
Gross/Histopathological and Microbiological Evaluation
Each grasscutter was euthanized with diethyl ether on developing an undermined ulcer. The carcass was exsanguinated by cardiac puncture and examined for gross and histopathological changes, presence of AFBs,
Statistical analysis was conducted with the Statistical Package for the Social Sciences (SPSS), Standard version, Release 12.0.1. (SPSS Inc. 1989-2003). The case summaries procedure was used to determine the mean, standard deviation, minimum and maximum values of the times of onset of lesions for each inoculum dose (experimental group). After conducting the above-mentioned determinations the data were screened for normality and homogeneity of variance. Most of the data violated the assumptions of the normal distribution and therefore were analysed with nonparametric tests. The Mann-Whitney test was used to determine if the time of lesion onset varied with inoculum dose (McFarland 1 and 5). Spearman's rho was used to (i) determine if there was any association between the inoculum dose and time of lesion onset and (ii) to determine the direction, strength and significance of the associations. A P value less than 0.05 was considered statistically significant.
All the animals in groups 1 (McFarland 1) and 2 (McFarland 5) developed similar clinical lesions regardless of the inoculum dose. However, the higher the inoculum dose, the earlier the onset of the lesions and the more severe their presentation. Each of the grasscutters in groups 1 and 2 presented a pinpointed or broad erythematous lesion at the site of inoculation 24-48 hours after inoculation however, no lesion was observed in any of the animals in group 3 after inoculation (hereafter ‘after inoculation' is abbreviated to ai). In both experimental groups the erythematous lesion was replaced by a scale (4-7 days ai), then one or two papules (9-19 days ai), which were sited 2.2-3.7.cm from the point of inoculation. Thereafter the infection progressed a little differently with each group. In group 1 the papule disappeared and was replaced by a blister-like lesion (Figure 1) (33-47 days ai), represented by a small skin elevation without any palpable content. The ‘blister' widened, hardened and became a nodule (Figure 2) (83-91 days ai), which disappeared and was replaced by a scratch (133-151 days ai). Exudate from the scratch hardened into a crust, followed by the formation of a small scab (135-154 days ai). In group 2, the papule developed into a nodule (29-37 days ai), devoid of the ‘blister' stage'. A superficial triangular sore (Figure 2) developed (35-43 days ai) on the central zone of the nodule, healed spontaneously within 5-7 days and resulted in the development of a scab (Figure 3) (40-49 days ai). The scabs from both experimental groups progressed similarly; the scabs adhered tightly to the skin, hardened and became brittle with time, which upon examination were found to contain AFBs. With the development of the scab the infection stalled, during this phase the scabs either persisted or disappeared intermittently such that both groups of animals often remained without any visible sign of infection for long periods. On the whole, this indeterminate or inactive infection phase was of a longer duration in the animals in group 1 (269-371 days) than those in group 2 (265-283 days); the differences were however not statistically significant (P>0.05). Surprisingly, with 2 of the animals in group 1 the infection did not progress beyond the scab stage (inactive infection) during the course of the study. The rest of the animals in both experimental groups relapsed after the long inactive infection phase and experienced a fulminant progression of the infection. Briefly, the inactive infection ended with sudden congestion of the thigh and portions of the pelvis as follows: (i) with reference to the animals in Group 1, the congestion initially localized around the scabs or the lower half of the thigh (in the absence of a scab) and subsequently engulfed the entire inoculated thigh. The congested area swelled slightly and was accompanied by gradual elevation of the scabs or the softening of certain congested areas (in the absence of a scab). The scabs and the softened areas eroded from underneath with the onset of purulent inflammation and eventually culminated in the formation of an undermined ulcer (441-540 days ai) with abundant fatty tissue on its floor. In the case of one of the animals in group 1, the entire pelvis, scrotum and contralateral limb became congested and oedematous (pitting oedema); the animal died a day after the onset of the oedema; (ii) regarding the animals in Group 2, the entire inoculated limb (i.e. thigh to the footpad) became slightly or intensely congested and oedematous ((Figure 4) and in some cases extended to the thigh of the contralateral limb. With 3 of the animals (2 females and 1 male) the congestion and oedema were extensive such that the entire pelvis and contralateral limb were severely affected within 48 hours, necessitating their immediate euthanasia (although they had not reached the research endpoint). Regarding the rest of the animals in group 2, the oedematous limbs (inoculated and non-inoculated) lacerated spontaneously with the subsequent development of a large scab. The scab eroded from underneath (Figure 5) with the onset of purulent inflammation, which culminated in the formation of an undermined ulcer (Figure 6) (368-379 days ai). Unexpectedly, 2 of the grasscutters with undermined ulcers (one presented the undermined ulcer on the inoculated limb while the other presented it on the contralateral limb) showed signs of spontaneous healing before the scheduled date of euthanasia, and therefore were left for further observation. Strangely the ulcers on both animals healed completely within 28-36 days without any evidence that the animals had previously presented undermined ulcers or any other BU lesion. None of the animals in the control group developed any visible or palpable lesion throughout the period of observation.
The differences in the onset of the lesions between the two experimental groups were statistically significant with respect to the development of scale (P<0.01), papule (P<0.01), nodule (P<0.01), superficial sore/scratch (P<0.01), scab (P<0.01), relapse (P<0.05) and undermined ulcer (P<0.05). There was negative correlation between the inoculum dose and time of lesion development: erythema (-.229, P>0.05), scale (-.872, P<0.01), papule (-.873, P<0.01), nodule (-.872, P<0.01), superficial sore/scratch (-.869, P<0.01), scab (-.869, P<0.01), inactive infection (-.252, P>0.05), relapse (-.521, P<0.05) and undermined ulcer (-.521, P<0.05). The descriptive statistics are presented in table 1.
The microbiological findings were similar in the two experimental groups. The blood, urine and faeces were consistently AFB negative. The exudate and pus from the inoculated/diseased limbs were consistently AFB positive while similar samples from their non-inoculated/diseased limbs were consistently AFB negative.
Grossly, the skin and subcutaneous tissue on the rump, thighs and lower limbs of all the grasscutters from the two experimental groups looked slightly or intensely congested. However that of the 4 grasscutters (from groups 1 and 2) with pronounced oedema looked cooked (coagulative necrosis) and the epidermis was tightly bound to the subcutaneous tissue, which was also calcified. In view of the pronounced oedema, fluid oozed from the subcutaneous tissues during dissection. Surprisingly, though the fluid from the inoculated/ diseased limbs was AFB positive in all four grasscutters, the fluid from their non-inoculated/diseased limbs was AFB negative. The exudate from both limbs in all four grasscutters also contained
Histologically, the epidermis in both the inoculated/diseased and non-inoculated/diseased sites of all the experimental animals was generally intact but the dermis and subcutaneous tissue were oedematous, inflamed and necrotized (Figure 7a), and occasionally contained inflamed and occluded blood vessels. The skin at the original sites of inoculation were without AFBs, however, the skin where the infection eventually manifested in the inoculated limb was laden with clusters of extracellular AFBs (Figure 7b); while the skin of the non-inoculated/ diseased limb was AFB negative. A few AFBs were detected in blood vessels of detached epidermises (Figure 8a) and necrotized areas of the dermis (Figure 8b). The scabs covering the ulcers were in various stages of repair represented by vascular granulation tissue, fibrous granulation tissue and fibrous scar tissue. The young granulation tissue was laden with AFBs (Figure 9a) while the fibrous scar tissue was sparingly embedded with AFBs (Figure 9b).
One other animal in Group 1 presented a large granulomatous lymph node (weighed 1.57g while normal lymph nodes weighed 0.04g – 0.06g) in the subcutaneous tissue on the spinal column (Figure 10a). Histologically, the granulomatous lymph node consisted of peripheral scar tissue, blood vessels, granulomas (Figure 10b), mixture of lymphocytes, neutrophils, epithelioid and giant cells, and several extracellular AFBs. The rest of the animals also presented a few chronically inflamed and necrotized subcutaneous lymph nodes (0.25g-0.28g) that contained AFBs.
At necropsy it was detected that all the animals had contracture (Figure 11a) in one or both hind limbs. The affected limbs were occasionally smaller in volume by 2.1-3.5cm and shorter in length by 1.6-2.0cm than the unaffected limbs. The sciatic nerves in the inoculated limbs were thicker than the sciatic nerves in the non-inoculated limb by 0.35-0.37 cm. The popliteal lymph nodes in the inoculated/diseased limbs were also larger than the popliteal lymph nodes in the non-inoculated/healthy limbs by 0.51- 0.52g. Though the popliteal lymph nodes were inflamed and necrotized they contained no AFBs. Histologically the muscle fibres in parts of the contracted limbs were curvilinear (Figure 11b) but other findings such as myolysis, myositis and neurogenic muscular atrophy were common to all the other limbs.
Histologically, the muscles of both the inoculated/diseased and non-inoculated/ diseased limbs were similarly affected and the most prominent and diffuse lesions were neurogenic muscular atrophy, coagulative necrosis and myositis. Neurogenic muscular atrophy (Figure 12a) was represented by groups of angular atrophied muscle fibres, sometimes accompanied by perineuritis and perivasculitis in the underlying connective tissue (Figure 12b). The muscle fibres were extensively necrotized (coagulative necrosis) in some areas resulting in the dissolution of the fibres, thereby giving them a lacy or moth-eaten appearance (Figure 13a); the vacant areas were taken up by scarring tissue and occasionally by fibrofatty tissues. Some areas in the muscles were diffusely oedematous; other areas, including the interstitium were chronically inflamed and contained several epitheloid cells and a few giant cells, including Langhans giant cells. The underlying connective tissue was invaded with AFBs (Figure 13b) but no AFBs were detected in the muscle or nerve fibres.
The bones in both the inoculated/diseased and non-inoculated/diseased limbs were diffusely and chronically inflamed and necrotized. Osteomyelitis (Figure 14a) was advanced and was represented by chronically inflamed and necrotized bone marrow accompanied by thinned, broken, eroded and in some cases deformed trabeculae. Though the affected bones were filled with pus cells and several giant cells (Figure 14b) they were unexpectedly completely devoid of AFBs.
A few (2-5) AFBs were detected in smears of the small intestines and caeca of all the animals in the two experimental groups; however none were detected in the histological slides. All other tissues (brain, heart, lung, liver, spleen, pancreas, kidney, urinary bladder, testis, scrotum, uteri, fallopian tubes, ovaries, sciatic nerves), which were examined by smear, culture and histology were without AFBs and fast-growing organisms.
The animals in the two experimental groups presented similar clinical lesions, but with obvious differences in the incubation periods and severity of lesions; the higher the inoculum dose, the earlier the onset of the lesions, and the more severe their presentation. These findings confirm observations of a mouse study (Addo et al, 2005) but contradict those of Walsh
Clinically, BU in humans manifests as papule, plaque, nodule, oedema (usually non-pitting), undermined ulcer, osteomyelitis and contracture (Thangaraj et al, 1999; WHO, 2001a & 2001b), which are histologically accompanied by extracellular acid fast bacilli (AFBs). The grasscutters manifested all but one (plaque) of the presentations, and the lesions were accompanied by extracellular AFBs, suggesting that grasscutters could be good BU animal models. The grasscutter seems to be a better BU animal model than the mouse (the most used BU animal model) because it elicited a greater number of lesions and its larger body size also eased the evaluation of lesions. Regrettably it took a long time for the lesions to manifest, a finding that detracts from its suitability. Notwithstanding this seeming drawback, the long incubation period may possibly be a truer reflection of what actually occurs in humans.
One obvious clinical difference between the grasscutter model and BU patients was with the presentation of the 'blister'-like lesion. A ‘blister' has never been reported as a BU lesion therefore we have difficulty explaining its development in the grasscutter. That notwithstanding, its presentation by only animals in group 1 (group of animals that received the lower inoculum dose) and its unequivocal absence in animals in group 2 (group of animals that received the higher inoculum dose) suggests that probably most BU patients are infected with higher doses and as such do not develop ‘blisters' in the course of their infection.
BU oedema is usually non-pitting (WHO, 2001a & 2001b), whilst that in the grasscutter was usually the pitting type, suggesting that the grasscutter model may only offer limited information on the pathogenesis of oedematous BU lesions. Oedema in the grasscutter was not only confined to the inoculated limb but extended to the contralateral limb, rump, tail and scrotum (hydrocel). Notably, these non-inoculated areas were conspicuously devoid of AFBs, suggesting that the oedema might be attributable to the systemic effect of the toxin; confirming similar findings and assertion in a mouse BU study (Addo et al, 2005) and also lending support to the suggestion that oedematous BU lesions involving large body areas are attributable to a systemic effect of the toxin (WHO, 2001b),
It is also reported that sometimes the oedematous form of the disease progresses rapidly and is associated with a worse prognosis (Thangaraj et al, 1999; WHO, 2001b). This was confirmed in the grasscutter study, given that the grasscutters progressed from an inactive infection to a fulminant form of the disease with the onset of diffuse oedema which in some of the cases resulted in the death of one grasscutter and necessitated the euthanasia of three others.
Clinically, all the grasscutters healed spontaneously in the course of their infection, confirming reports that some BU patients at some stage of the infection heal spontaneously by an unknown mechanism (Thangaraj et al, 1999; Walsh et al, 1999; WHO, 2001b; Johnson, 2005). Healing in BU patients is represented histologically by granulomatous infiltration, granulation tissue, fibrosis and a depressed scar with very few or no AFBs (WHO, 2001b). Similar histological findings were associated with the scabs that covered the undermined ulcers; however, unlike the case with BU patients, AFBs were consistently present in the scab. The presence of AFBs in the friable scabs during the period of inactive infection was unexpected and worrying because the finding suggests that ‘healed' lesions of BU patients should be handled with circumspection since they may be infectious. Notwithstanding that the AFB-laden scabs were not cultured to ascertain the viability of the organisms, it would be safer to err on the side of caution, than to be infected for being over optimistic, especially since the mode of transmission is unknown.
Recurrence of infection occurs in about 16-47% of healed BU patients (WHO, 2001a, 2001b; Teelken et al, 2003). In our study, 90% of the spontaneously healed grasscutters relapsed and it was always a prelude to a more severe infection, which always culminated in the development of the undermined ulcer. This finding suggests that the microbes were actually disseminating during the ‘healed' phase, therefore BU patients have to be maintained on systemic antibiotic treatment during the ‘healed' phase to prevent a recurrence. In view of the non-availability of an effective antibiotic, the currently recommended WHO drugs could be tried in the interim; the duration of the treatment would also have to be determined. Among the animals that relapsed, 2 healed spontaneously again after having developed undermined ulcers. The reason for this occurrence may not be immediately evident; however, in view of the previous healing incidents and relapses the possibility exists that the animals could relapse again. As observed in this and the mouse study (Addo et al, 2005), the grasscutters and mice that relapsed with concomitant metastasis had earlier presented clean healed scars. This situation goes to underscore the urgency of identifying antimicrobials with bactericidal effect.
For the most part, all the BU lesions in the grasscutters developed away from the site of inoculation, specifically towards the posterior (rump, tail, scrotum) and lower extremities (distal portion of thigh, entire hindlimb to the level of the footpad and entire contralateral limb) confirming the organism's preference for cooler areas (Thangaraj et al, 1999) and also providing evidence that by virtue of its toxin it has the ability to disseminate widely. The findings of our study also suggest that dissemination probably occurred by lymphatic and haematogenous spread because AFBs were detected in both the lymph nodes and blood vessels. These findings in the grasscutters further suggest that the site(s) at which BU lesions manifest in humans may not necessarily be the site of entry for the bacterium but may rather be the preferred site(s) of the bacterium and its toxin. This assertion is plausible because there are several diseases that manifest at sites far removed from their points of entry (Cotran et al, 1999). Chicken pox and measles viruses for example enter through the airways but manifests themselves first as skin rashes (Cotran et al, 1999). If this possibility is borne in mind in our quest to unravel the mode of
It is reported that BU lesions rarely extend into the underlying muscle (WHO 2001B) because it is speculated that muscles slow down the disease process, though not immune from attack by
Though coagulative necrosis and inflammation are not unusual BU findings; the consistent detection of neurogenic atrophy and perineuritis are enormously important. Clinically the pathology was manifested as a decrease in muscle volume and contracture; symptoms that are also present in BU patients. The manifestation of neurogenic atrophy suggests that BU affects the peripheral nerves, which in turn lead to the destruction of the muscle fibres (microscopic atrophy, clinical contracture) and could be the reason for the absence of pain at some stages of the disease. This finding also goes to confirm that as BU progresses all elements of the skin are affected, including the nerves (Thangaraj et al, 1999). In view of the nerve involvement BU should be studied in the light of what is known about leprosy. Rather frighteningly, these neurogenic lesions extended to the contraletaral limb in the absence of
The lesions in the bones were similar in all respects to those of BU patients. However, a surprising and important difference was the unequivocal absence of AFBs in the grasscutter lesions. AFBs are present in human BU lesions (WHO, 2001b) and were also observed in a mouse BU study (Addo et al, 2005). It has been assumed that BU bone lesions are the result of haematogenous spread of
In BU patients, 50% of the osteomyelitic lesions are co-infected by pyogenic organisms such as streptococci, staphylococci and corynebacterium sp. (WHO, 2001b). The osteomyelitic lesions in the grasscutters were similarly infected by a number of fast-growing organisms; a finding that sheds further light on why BU is difficult to treat. The involvement of these microbes further complicates the treatment of BU because it implies that anti-BU drugs should have broadspectrum activity to be really effective or other antibiotics have to be administered alongside the anti-BU drugs in order to effectively treat the disease.
The lesions in the lymph nodes were similar to those found in BU patients. Similarly, AFBs were only detected in lymph nodes in the vicinity of the skin lesions but never in the regional nodes. Granulomatous changes are also not usually seen in BU patients (WHO, 2001b); similarly, only one granulomatous lymph node was detected in the study.
The presentation of hydrocel by the grasscutters confirms similar findings in BU mice (Addo et al, 2005) and reports of genital involvement (WHO, 2001a). The hydrocel was devoid of AFBs suggesting that it was due to the systemic effect of
In conclusion, the clinical and histopathological presentations of
The study was funded by the Noguchi Memorial Institute for Medical Research. The authors thank Messrs David Appiah, Alexander Kemausuor and Isaac Hudson-Odoi for technical assistance.
Phyllis Addo (DVM, PHD) Department of Animal Experimentation, Noguchi Memorial Institute for Medical Research College of Health Sciences, University of Ghana, P. O. Box LG 581, Legon. Accra, Ghana Email: email@example.com firstname.lastname@example.org Fax 233 21 502182 Telephone: 233 21 501178/9