With a maximal length of 1 mm, Tunga penetrans, the so-called chigoe- or jigger-flea (Linné 1758; syn. Sarcopssylla, Dermatophilis, Rynchoprion), is the smallest flea known (Connor 1976). True to its species name, the female burrows into the skin of its host. There it remains for a period of up to 5 weeks, during which it matures, produces and releases eggs and finally dies. Within 1-2 weeks, the flea increases its volume by a factor of roughly 2,000-3,000, frequently reaching a diameter of up to 1 cm (Geigy and Herbig 1949; Linardi 2000; Rey 1992). The entire parasite remains completely buried in the epidermis, with the exception of the posterior parts of the abdomen bearing the anus, the genital opening and four pairs of large stigmata. The protruding rear end leaves a sore, 240-500 µm in diameter, which may serve as an entry point for pathogenic micro-organisms (Feldmeier et al. 2002). In humans, T. penetrans lesions can be found at any part of the body, although the periungual regions of the toes are preferential penetration sites (Heukelbach et al. 2002).

Although the first scientific description of tungiasis dates back to the sixteenth century (Guerra 1968), systematic clinical and pathological studies do not exist. Geigy and Herbig (1949) described the morphological and structural changes in fleas collected from patients in the French Congo, the Belgian Congo and Puerto Rico and proposed to divide the development of the female flea into four stages. As the exact time of penetration was not known and clinical findings were not documented, the authors could not conclude on the duration of the stages nor correlate morphological findings to clinical observations. Also, they did not have specimens of fleas in the involution phase.

Many case reports exist, offering clinical aspects of T. penetrans in travellers returning from tropical Africa, the Caribbean and South America, but without describing the morphology of the parasite. In most cases, these patients had only a single lesion, so that staging was not possible (for a review, see Franck et al. 2002).

In the present paper, we attempt to combine observations from a comprehensive clinical study with the results of systematic histological sectioning and scanning electron microscopy (SEM) of specimens obtained from the same patients, in order to elucidate the relationship between morphological changes of the parasite, the histopathological appearance of the lesion and the ensuing clinical picture. Based on these findings, we propose the "Fortaleza classification", a staging system which takes into account the dynamic nature of tungiasis and which allows the division of this peculiar ectoparasitosis into five consecutive stages.

The so-called chigoe- or jigger-flea, Tunga penetrans (Linné 1758; syn. Sarcopsylla, Dermatophilis, Rhynchoprion), is a fascinating example of an arthropod which, having started from an ectoparasitic life-style, has adapted to a peculiar type of endoparasitism for the reproductive phase of its life cycle, which takes place in various domestic, peridomestic and wild animals and in humans (Cooper 1967; Rietschel 1989; Vaz and Rocha 1946). Once the female flea penetrates into the epidermis of its host, a complex sequence of structural and morphological changes is initiated. A period of amazing hypertrophy is followed by a phase of involution, at the end of which the parasite dies and eventually is eliminated by repair mechanisms of the skin, as occurs with other types of foreign body. Thus, in the vertebrate host, the natural history of T. penetrans is a dynamic, albeit self-limiting process. In this study, the natural history of tungiasis in man is elucidated and a staging system applicable to both clinical and epidemiological research is proposed.

As with other dynamic diseases, staging is a prerequisite for various types of investigations. It permits classification of clinical manifestations, allows a more precise diagnosis, enables the assessment of therapy and facilitates the evaluation of control measures. Obviously, if the dynamic nature of tungiasis is not taken into consideration, the wrong conclusions are easily drawn. For example, various chemotherapeutic approaches to kill embedded fleas were undertaken with niridazole, thiabendazole and ivermectin; and the therapeutic efficacy of these drugs was always highly acclaimed (Ade-Serrano et al. 1982; Cardoso 1981; Saraceno et al. 1999). However, those authors completely neglected the fact that T. penetrans will die anyway and eventually all lesions will heal, even without specific interventions. Thus, the assessment of chemotherapy of tungiasis necessitates the understanding of the natural history of the lesions, and efficacious treatment can only mean that the normal duration of the disease is abrogated. This, in turn, requires knowledge of how long the different stages normally last. Similarly, the diverging clinical pictures reported to occur in patients returning from a holiday in the tropics and the different diagnostic guidelines provided by various authors are simply explained by the fact that patients were examined at various stages of the disease (Franck et al. 2003).

So far, only Geigy and Herbig (1949) have attempted to describe the natural history of tungiasis in man. However, their work is characterised by several shortcomings. First, the description of structural and morphological changes was based only on the microscopic examination of fleas conserved in Dubosque's preservative. Second, as the time of evolution of the lesions was not known, the authors could not confirm the duration of the stages they described. Finally, they only examined specimens taken during the hypertrophy phase and thus neglected later stages during which the lesion regresses and eventually the remnants of the parasite are sloughed.

In the present paper, we propose the Fortaleza classification, a staging system which divides the natural history of T. penetrans in man into five stages and which is based on a combination of morphological observations, histopathological findings and the clinical picture.

Stage 1, the penetration phase, is the shortest of the five consecutive stages. It usually takes only 3 h (ranging up to 7 hours), a time-span previously reported by Karsten (1865). Presumably, the time necessary for penetration depends on the texture and particularly the thickness of the keratin layer. It can be shorter when penetration occurs in the neighbourhood of already embedded fleas, where the inflammation surrounding existent lesions has softened the skin. In rats, in which T. penetrans usually penetrates the footpad or the plantar part of the toes, penetration seems to require more time than in man (Heukelbach, unpublished data).

Nothing is known as to whether the penetration is exclusively a physical process in which the flea uses its compact saw-like proboscis to drill a hole in the epidermis, enabling the ovoid body to penetrate the keratin layer or whether this is facilitated by enzymes secreted through its salivary glands. The observation that penetration is frequently accompanied by itching and erythema is a hint that at least vasoactive substances are released by the penetrating flea. Also, it is assumed that substances are released which have an anticoagulant effect, impeding the clotting of blood during feeding.

There is considerable difference between individuals in the perception of penetration. This may be due to differences in the sensory sensibility of topographic sites affected or, if an allergic pathomechanism is assumed, it may depend on previous sensitisation. Karsten (1865) suggested that the perception of pain during penetration simply depended on the experience an individual had with tungiasis: people living in the endemic area easily recognise the pain and itching accompanying penetration and differentiate these sensory perceptions from, e.g. the bite of a midge, whereas "newcomers" tend to ignore these symptoms. Whenever symptoms were perceived, a particular type of itching ("coçeira boa": pleasant itching) seemed to predominate (Karsten 1865).

The beginning of the separation of abdominal segments two and three and the generation of intersegmental skin is accompanied by functional changes. As Geigy and Herbig (1949) pointed out, the penetrating flea already feeds on blood. This was confirmed in our study by iron staining of histological sections which showed the presence of intact and partially digested erythrocytes in stage 1. As the majority of the erythrocytes were still intact, it is improbable that the red blood cells were residues of previous temporary feeding during the free-living stage and confirms the assumption that anticoagulant factors are excreted.

In stage 2, the head of the flea reaches the stratum lucidum and the proboscis is firmly inserted in sub-epidermal tissue. This stage is also characterised by a rapid increase in the hypertrophy zone between abdominal segments two and three. This zone forms a bulge, which, when seen transversally, appears similar to a life belt (Fig. 4). As observed by Schimkewitsch (1884), the rear end, the genital opening and the four pairs of stigmata form a cone protruding through the skin, thereby connecting the surface of the skin with its deeper layers. It is therefore not surprising that bacteria were found at the host-parasite interface during this early stage of development of the neosome. These bacteria may have been on the outer surface of free-living fleas and were hence passively carried into the epidermis; or they may have been actively introduced from microcolonies present on the skin through scratching (Feldmeier et al. 2002b). Karsten (1865) observed that local inflammation was particularly intense when lesions were squeezed or scratched. That local inflammation increases between stages 1 and 2 is supported by an observation from Geigy and Herbig (1949). In stage 2, those authors observed that the gut of the flea contained many leukocytes beside the expected erythrocytes. This corroborates our findings that, in histological sections of stage 2 lesions, neutrophils and lymphocytes formed perivascular infiltrates and seemed to migrate to the host-parasite interface.

Intense infection and subsequent inflammation may not only be caused by the availability of a steadily growing inert surface on which bacteria can form biofilms (Feldmeier et al. 2002b). Bacterial propagation may also be due to an altered quality of the surface. Geigy and Herbig (1949) observed that the outer membrane of newly built intersegmental skin has many tiny grooves. Supposedly, both in these grooves and in the niches formed at the junction between chitin segments and intersegmental skin, bacteria can easily adhere and multiply (Fig. 7).

In stage 3, the morphological changes become impressive. In substage 3a, the hypertrophy zone develops into a sphere with a diameter of up to 10 mm. Simultaneously, the three parts of the second abdominal segment are stretched and bent apart, so that the flea gives the impression of a three-leafed clover when seen from the cranial direction (Fig. 8). The increasing hypertrophy zone correlates with peculiar clinical findings: the balloon-like intersegmental skin shimmers through the keratin layer as a white circular halo. The firm consistency of the watch-glass-like protrusion presumably reflects the pressure that the increasing hypertrophy of the ovarian ductules, the gut, the Malpighian ductules and the muscles exert against the intersegmental skin (Geigy and Herbig 1949).

According to our definition, by substage 3b, the hypertrophy has reached its zenith. Morphologically, this corresponds to a steady size of the sphere and an increase in the thickness of the exoskeleton of the abdominal segments adjacent to the hypertrophy zone. This causes a kind of rim rampart which towers above the cone of the last two segments (Geigy and Herbig 1949). The mini-caldera seen in the skin clearly reflects these morphological changes (Fig. 13). As the stock of eggs in the ovaries is depleted, presumably the pressure exerted against the intersegmental skin decreases. The loss of the firm consistency of the lesion and its subsequent wrinkled appearance (Fig. 16) reflect the functional and structural changes taking place in substage 3b.

We observed five characteristics which point to stage 3 being the phase with highest metabolic activity: excretion of faeces, expulsion of eggs, a brownish-watery secretion, a pulsation phenomenon and vertical movements of the cone. These findings have never yet been described. First, previous investigations were limited to the microscopic examination of preserved material (Geigy and Herbig 1949) and, second, such characteristics are only observable in those cases where lesions are examined repeatedly and for a considerable period of time. Besides, being diagnostic clues for physicians not accustomed to seeing patients with tungiasis, these criteria help to assess the efficacy of chemotherapy.

The form (a helical tread) and the consistency (adhesive) of faeces excreted by T. penetrans are similar to faecal coils produced by T. monositus, a flea that parasizes exclusively the pinna of the ear of a species of mouse (Lavoipierre et al. 1979). In this animal, faeces is so adhesive that male fleas are frequently trapped when they try to mate with embedded females. As lesions almost invariably occur in clusters and faeces always spreads into the cutaneous grooves surrounding the cluster, it is tempting to speculate that faeces act as a kind of pheromone attracting males. Obviously, the odds are higher for a single male to inseminate several embedded females per unit of time if lesions are located near to each other and the male is directed to such clusters by a chemical attractant.

As adhesiveness of eggs is also known from flea species not penetrating their host, it is thought to be an advantage for the propagation of the species, because non-adhesive eggs would fall to the ground after they are laid (Jacobs and Renner 1988). The observation that embedded females expel their eggs with a considerable velocity may have an additional advantage. It avoids the chance that eggs could come into contact with faecal material spread in the dermal papillae around the rear cone. The clumsy faeces could impede the eggs in falling to the ground, and faecal material sticking to an egg could impair development of the larva.

The irregular expulsion of eggs has also been observed in other Tunga species (Lavoipierre et al. 1979).

The brownish watery secretion, the pulsation phenomenon and the vertical movements of the rear cone apparently belong together. Sand fleas obviously shed the watery components of the blood, thereby concentrating energy-rich leukocytes and erythrocytes. This phenomenon is also known from other blood-sucking insects (Mehlhorn 2001a, 2001b).

Stage 4 is a fluid transition between a further involution of the neosome and elimination of the remnants of the dead flea by repair mechanisms of the skin. The sequence of events is clearly demonstrated by the histological observations made in the present study. Whereas in substage 4a the outer surface of the parasite becomes heavily covered with neutrophils, eosinophils and lymphocytes, in substage 4b these cells invade the carcass of the flea. The loss of reproductive activity is evidenced by the fact that fewer and fewer eggs are expelled and that the production of faeces, the watery secretion and the pulsations eventually stop completely. Macroscopically, the lesion appears to dry out and finally is covered with a black-brownish crust of necrotic material (Fig. 18). It remains unclear whether the discoloration observed in substage 4b is an indicator of the ageing of the parasite's intersegmental skin, or occurs due to faeces or clotted blood spread at the host-parasite interface.

According to our clinical observation, the natural history of tungiasis in man does not end with the elimination of the carcass of the parasite. For several months, a characteristic residue remains in the skin which looks like a punched-out circular depression (Fig. 19). Such residues are particularly frequent at sites with a thick keratin layer, such as the heels and the sole. Histologically, these residues correspond to small superficial cutaneous scars. The residues of stage 5 provide a record for assessing the tungiasis history of an individual (Fig.20). Hence, they could be used to evaluate the efficacy of control measures, if the ratio of stages 2-4 to stage 5 lesions are determined before and after intervention at the population level. Other sequels of previous tungiasis, such as deformation of toenails, or loss of nails and digits are less constant and mainly occur in patients with extremely severe disease (Feldmeier et al. 2002a).

Although the periungual areas of the toes and especially the nail rim (Figs. 5, 6) are clearly preferred sites, tungiasis may occur at any part of the body, including the fingers and the genitals (Heukelbach et al. 2001, 2002). However, we could not find differences in the natural history of tungiasis in relation to lesions occurring at different topographic sites.

An important finding of this study is the good correlation between clinical and histopathological findings. Whenever macroscopically inflammation was present as indicated by erythema, oedema, warmth and pain in histological sections, vasodilatation, a cellular infiltrate and (occasionally) spongiosis were observed (data not shown). Also, hyperkeratosis frequently coincided with macroscopically observable desquamation of the skin and parakeratosis. However, we could not identify a particular histological pattern specifically associated with any stage of the disease. As biopsies were frequently taken from a cluster of lesions or from a site where presumably lesions had existed in the past, we assume that the histological picture observed in our patients in fact mirrored a mixture of tissue-reaction patterns surrounding both early- and late-stage lesions as well as already eliminated neosomes.

Taken together, the present paper describes for the first time the complete natural history of T. penetrans in man from the beginning of penetration to the residual stage. It is shown that the sequence, duration and histological patterns of evolving and regressive phases of the female flea are substantially different from other Tungidae studied, such as T. monositus, an ectoparasite with extremely specific parasite-host relationship (Lavoipierre et al. 1979). The classification proposed integrates morphological, histopathological and macroscopic findings and seems valuable for various research purposes. It now has to be demonstrated that the Fortaleza classification is also valid to describe the natural history of T. penetrans in domestic animals, such as dogs and cats, as well as in rodents.