Portal de Periódicos da CAPES

Cerebrospinal fluid B lymphocyte identification for diagnosis and follow-up in human African trypanosomiasis in the field

2009; Wiley; Linguagem: Inglês

10.1111/j.1365-3156.2009.02400.x

1365-3156

Bernard Bouteille, Ghislain Armel Mpandzou, Raymond Cespuglio, Stéphane Ngampo, Rosanna W. Peeling, Philippe Vincendeau, Alain Buguet,

Research on Leishmaniasis Studies

Objectives In human African trypanosomiasis (HAT, sleeping sickness), staging of disease and treatment follow-up relies on white cell count in the cerebrospinal fluid (CSF). As B lymphocytes (CD19 positive cells) are not found in the CSF of healthy individuals but occur in neurological disorders such as multiple sclerosis, B lymphocyte count may be useful for field diagnosis/staging and therapeutic follow-up in HAT. Methods Seventy-one HAT patients were diagnosed and 50 were followed-up 6–24 months after treatment. White cell counts were used for conventional staging (stage 1, ≤5 cells/μl CSF, n = 42; stage 2, ≥20 cells/μl, n = 16) and intermediate stage (6–19 cells/μl, n = 13). Slides containing 1 μl of CSF mixed with Dynabeads® CD19 pan B were examined microscopically to detect B cell rosettes (bound to at least four beads). Results Stage 1 patients exhibited zero (n = 37) or one CSF rosette/μl (n = 5), contrary to most stage 2 patients (14/16: ≥2 rosettes/μl). Intermediate stage patients expressed 0 (n = 9), 1 (n = 3) or 2 (n = 1) rosettes/μl of CSF. During follow-up, rosette counts correlated with white cell count staging but were much easier to read. Conclusion B cell rosettes being easily detected in the CSF in field conditions may be proposed to replace white cell count for defining HAT stages 1 and 2 and limit uncertainty in treatment decision in patients with intermediate stage. Identification des lymphocytes B dans le liquide céphalo-rachidien pour le diagnostic et le suivi de la trypanosomiase humaine africaine sur le terrain Objectifs: Dans la trypanosomiase humaine africaine (THA, maladie du sommeil), les stades de la maladie et le suivi du traitement s’appuient sur le nombre de globules blancs dans le liquide céphalo-rachidien (LCR). Comme les lymphocytes B (cellules CD19 positives) ne sont pas présents dans le LCR des individus en bonne santé mais surviennent dans les troubles neurologiques comme la sclérose en plaques, la numération des lymphocytes B pourrait être utile pour le diagnostic/classification sur le terrain et le suivi thérapeutique de la THA. Méthodes: 71 patients HAT ont été diagnostiqués et 50 ont été suivis durant 12 à 24 mois après le traitement. La numération des globules blancs a été utilisée pour la classification classique (stade 1, ≤ 5 cellules/ml CSF, n = 42; stade 2, ≥ 20 cellules/ml, n = 16 et stade intermédiaire 6 à 19 cellules/ml, n = 13). Des lames contenant 1 ml de LCR mélangé avec du Dynabeads® pan B CD19 ont été examinées au microscope afin de détecter les rosettes de cellules B (liées à au moins 4 billes). Résultats: Les patients au stade 1 présentaient zéro (n = 37) ou une rosette/ml de LCR (n = 5), contrairement à la plupart des patients au stade 2 (14/16: ≥ 2 rosettes/ml). Les patients au stade intermédiaire exprimaient 0 (n = 10), 1 (n = 2) ou 2 (n = 1) Rosettes/ml de LCR. Au cours du suivi, le nombre des rosettes corrélait avec la teneur en globules blancs, mais étaient beaucoup plus facile à lire. Conclusion: Les rosettes de cellules B facilement détectées dans le LCR dans des conditions de terrain peuvent être proposées pour remplacer la numération des globules blancs pour définir les stades 1 et 2 de la THA et limiter l’incertitude dans la décision de traitement chez les patients au stade intermédiaire. Identificación de linfocitos B en líquido cefalorraquídeo para el diagnóstico y seguimiento en campo de la tripanosomiasis humana Africana. Objetivos: En la tripanosomiasis humana Africana (THA, enfermedad del sueño), conocer el estadío de la enfermedad y el seguimiento del tratamiento dependen del conteo de leucocitos en líquido cefalorraquídeo (LCR). Puesto que los linfocitos B (CD19 + ) no se encuentran en el LCR de individuos sanos pero si en el caso de desórdenes neurológicos tales como la esclerosis múltiple, el conteo de linfocitos B podría ser útil para el diagnóstico en campo/estudio del estadío y seguimiento de la THA. Métodos: Se diagnosticaron 71 pacientes con THA y se realizó el seguimiento a 50, de 12 a 24 meses después del tratamiento. Se utilizaron los conteos de leucocitos para determinar de forma convencional el estadío de la enfermedad (estadío 1, ≤5 células/μL CSF, n = 42; estadío 2, ≥20 células/μL, n = 16) y estadíos intermedios (6-19 células/μL, n = 13). Las láminas con 1 μL de LCR mezclado con Dynabeads® CD19 pan B fueron examinadas bajo microscopio para detectar rosetas de células B (unidas al menos a 4 perlas). Resultados: Pacientes en el estadío 1 tenían cero (n = 37) o una roseta de LCR /μL (n = 5), al contrario de la mayoría de pacientes en el estadío 2 (14/16: ≥ 2 rosetas/μL). Los pacientes en un estadío intermedio presentaban 0 (n = 10), 1 (n = 2) o 2 (n = 1) rosetas/μL de LCR. Durante el seguimiento, los conteos de rosetas daban resultados que se correlacionaban con los obtenidos mediante el conteo de leucocitos, pero eran mucho más fáciles de leer. Conclusión: Las rosetas de células B son más fáciles de detectar en LCR bajo condiciones de campo y por lo tanto podrían proponerse para reemplazar el conteo de leucocitos a lo hora de definir los estadíos 1 y 2 de THA y limitar la falta de certeza a la hora de tomar decisiones sobre el tratamiento de pacientes con un estadío intermedio. Human African trypanosomiasis (HAT) in western and central Africa is due to the transmission of Trypanosoma brucei (T. b.) gambiense by tsetse flies. The illness evolves from haemolymphatic stage 1 to meningoencephalitic stage 2. At stage 2, trypanosomes and/or mononuclear inflammatory cells appear in the cerebrospinal fluid (CSF), indicative of central nervous system (CNS) invasion. Untreated HAT evolves towards death in months or years. Pentamidine is successful in treating stage 1. Stage 2 is treated with melarsoprol or eflornithine which are toxic and difficult to administer (Bouteille et al. 2003). Melarsoprol provokes reactive arsenical encephalopathy in 2–10% of the patients of whom 50–75% die. Eflornithine is infused intravenously every 6 h for 14 days. A large study recently published by Médecins sans Frontières recommends the use of eflornithine rather than that of melarsoprol, especially in areas where melarsoprol effectiveness is low (Balasegaram et al. 2009). However, eflornithine may still be toxic (Balasegaram et al. 2009) and remain ineffective in T. b. rhodesiense HAT (Fèvre et al. 2008; Kennedy 2008). Its activity on HIV-infected patients is still under debate since the report by Pépin et al. (1992). Appropriate treatment of patients therefore requires the accurate determination of the stage of the disease (Chappuis et al. 2005; Kennedy 2008), which remains inconsistent and unsatisfactory (Kennedy 2008). The occurrence of trypanosomes in the CSF represents the only specific diagnostic test. However, tracking the infiltration of brain tissues by trypanosomes lacks sensitivity (Chappuis et al. 2005), as do clinical signs, blood tests and CSF cell counts. Broden and Rhodain (1908) considered a lymphocyte concentration of 5 cells/μl of CSF as the upper limit of CNS integrity, and attributed a large number of lymphocytes to the meningeal reaction. Joyeux and Sicé (1937) considered that the quality of CSF cellular elements is more relevant than their concentration. However, little progress has been made to identify CSF cellular types in HAT. The World Health Organization (WHO) recommends that the positive diagnosis of HAT relies on the microscopic detection of trypanosomes in lymph nodes and/or blood samples, while staging of the infection is based on the microscopic observation of trypanosomes and/or white cell counts in the CSF (WHO 2007). Conventionally, CSF cell counts of ≤5/μl and the absence of trypanosomes in the CSF correspond to stage 1. Stage 2 is characterized by the presence of trypanosomes and/or cell counts of ≥20/μl in the CSF. Between these thresholds (from 6 to 19 cells/μl) and in the absence of trypanosomes, an intermediate stage may be considered as some patients may be cured with stage 1 treatment while others will require stage 2 medications (Lejon et al. 2003; WHO 2007). Nevertheless, most national programmes use different cell count cut offs and do not consider the intermediate category, resulting in treatment decisions that may have adverse consequences to the patient. Although lacking specificity, the leucocyte count remains the key to a therapeutic decision, despite technical inconsistencies and difficulties in defining a threshold (Chappuis et al. 2005). The search for new specific biological markers for staging and therapeutic follow-up has encountered major difficulties to date (Lejon et al. 2008). Extensive polyclonal activation of B lymphocytes that proliferate and secrete high levels of immunoglobulins has been observed in HAT (Greenwood 1974; Kazyumba et al. 1986). Indirect immunofluorescence had revealed the presence of B lymphocytes in the CSF of stage 2 patients (Greenwood et al. 1976), which was recently confirmed by flow cytometry (Boda et al. 2009). In multiple sclerosis, the presence of B lymphocytes in the CSF, detected by flow cytometry, correlates with the degree of brain inflammation evidenced by magnetic resonance imagery (Kuenz et al. 2008). However, the determination of CNS involvement based on cytological changes in the CSF of HAT patients has not been explored in field conditions. To address this question, the present study aimed at evaluating the value of the presence of B lymphocytes in the CSF of patients with T. b. gambiense HAT in staging and post-treatment procedures. A rosette technique using a pan B cell marker, CD19 (Carter & Barrington 2004), was used to quantify the presence of B lymphocytes in the CSF. Seventy one patients were recruited from the Couloir focus on the right bank of the Congo River (Republic of Congo, Brazzaville) in June 2007 (group I), June 2008 (group II) and June 2009 (group III), and were included in the study. Patients gave informed consent, following a protocol approved by WHO Research Ethics Review Committee and authorized by the Congo Ministry of Health. In the 71 patients, HAT was confirmed by the presence of trypanosomes in the blood or lymph nodes, following WHO recommendations for clinical trials (WHO 2007). Clinical examination was carefully performed, with an emphasis on neurological signs as previously reported (Radomski et al. 1995). General symptoms, such as swollen lymph glands, isolated headaches, fever or fatigue, were regarded as signs of the ongoing infectious process. Neurological involvement was considered when at least two of the following symptoms were identified: sleep disturbances (nocturnal insomnia, daytime somnolence), sensory disturbances (uncomfortable diffuse superficial or deep sensations signing hyperpathia; pruritus with or without skin lesions), motor disturbances (abnormal gait, fine and diffuse tremor, myoclonic jerks), exaggerated deep tendon reflexes, presence of primitive reflexes (palmo-mental reflex, sucking reflex), loss of proprioceptive control (Romberg’s sign), and psychiatric manifestations (confusion, mood swings, agitation, aggressive behaviour, euphoria, absent gaze, mutism, indifference). Among the 71 patients, HAT stage determination was performed in 23 group I patients in July 2007, 27 group II patients in July 2008, and 21 group III patients in July 2009. Patients in group I were diagnosed and treated in July 2007 and followed-up after treatment in December 2007 (6-month follow-up), July 2008 (12-month follow-up), December 2008 (18-month follow-up), and July 2009 (24-month follow-up) using the same procedure. Group II patients diagnosed in July 2008 were followed-up 6 and 12 months after treatment. Group III patients were only included in the diagnosis group. Patients diagnosed at stage 2 were immediately treated with eflornithine or melarsoprol. Patients meeting the criteria for intermediate stage and stage 1 received pentamidine. During post-treatment follow-up, classification of group I and group II patients’ condition was adapted from a WHO experts report (WHO 2007; Table 2 and Figure 2). Compared to the initial health condition at the time of diagnosis, the classification included ‘favourable evolution’ and ‘completely cured’ (FE and CC: CSF cell count ≤5 cells/μl, no trypanosome in the CSF; a patient is considered to be completely cured after a 24-month follow-up without any relapse), ‘uncertain evolution’ (UE: biological CSF features of intermediate stage with CSF cell count between 6 and 19 cells/μl, or rising number of CSF leucocytes) and ‘still at stage 2 or suspected relapse at stage 2’ (R2: stage 2 CSF signs or CSF parasitological evidence or even worsening in clinical condition evoking stage 2). White cell counts/μl and CD19 pan B rosette counts/μl in the cerebrospinal fluid of 71 patients diagnosed for HAT in July 2007 (group I), July 2008 (group II) and July 2009 (group III) and in 50 patients (groups I and II) followed-up every 6 months during 12 (group II) or 24 months (group I) after treatment. The ordinate scale was changed to account for large values. Individual values are represented by small plain circles. The number of rosette count values at 0/μl is indicated below the CD19 B cells count column. The diagnosis procedure classified the patients as being at stage 1 (≤5 white cells/μl, no trypanosome), intermediate stage (6–19 white cells/μl, no trypanosome) or stage 2 (≥20 white cells/μl and/or trypanosomes). Follow-up 6 months after treatment was completed in 40 group I and group II patients. Follow-up 12 months after treatment was conducted in 38 group I and group II patients. Follow-up 18 months after treatment was conducted in 16 group I patients. Follow-up 24 months after treatment was conducted in 10 group I patients. The nine completely cured patients (CC, ≤5 white cells/μl CSF, no trypanosome, after a 24-month follow-up without any relapse) – including S52, S53, S66 and S68 – and one favourable evolution (FE, ≤5 white cells/μl CSF, no trypanosome) patient (S73) showed 0 rosette/μl of CSF. In the follow-up procedure, patients were classified as being in CC or FE, Uncertain evolution (UE: 6–19 white cells/μl, no trypanosome) or being still at stage 2 or showing evidence of relapse at stage 2 (R2, ≥20 white cells/μl and/or trypanosomes). Patients belonging to stage 2 and intermediate stage categories before treatment (at the time of diagnosis) and to UE and R2 post-treatment categories (during the follow-up) are indicated by their anonymous code (example: S74). At the 18-month follow-up examination, group I S73 relapsed at stage 2. Due to pregnancy 12 months after treatment, the patient had been exempted of lumbar puncture, following the recommendations of the Congo health authorities. Patient S73 was treated anew with melarsoprol and resumed normal CSF counts 6 months later at the 24-month check-up showing favourable evolution. Patient S74 also belonged to group I and was still showing criteria for uncertain evolution at the 6-month follow-up (18 white cells/μl; 10 rosettes/μl). The patient did not meet the 12-month follow-up appointment, but CSF cell and rosette counts were back to normal 18 months after treatment (4 white cells/μl; 1 rosette/μl). Patient S74 also missed the 24-month follow-up session. Patient S68’s condition improved at the 6-month (10 white cells/μl; 1 rosette/μl) and 12-month follow-ups (5 white cells/μl; 0 rosette/μl), worsened 18 months after treatment (10 white cells/μl; 3 rosettes/μl), and resumed normal CSF counts at the 24-month follow-up. Samples of CSF were obtained from patients at the time of HAT stage determination from a lumbar puncture and examined under microscopy for conventional (count of all cells without any cell identification) and rosette counts (specific count of B lymphocytes). The rosette count was performed as follows: a 500 μL CSF sample was mixed with Dynabeads® CD19 pan B (Dynal Biotech, Oslo, Norway) and agitated gently for 20 min. In order to validate the manufacturer’s recommendation regarding the use of KOVA® slides (Hycor Biomedical Inc., California, USA) and considering the preferential use of Nageotte slides by the field technicians, the initial diagnostic sample was placed on both Nageotte and disposable KOVA® slides, which are safe for technical use. Both samples were examined under direct light microscopy (400 ×) to search for trypanosomes, and establish leucocyte and rosette counts (cells to which at least four red beads are attached), indicative of B lymphocytes. Readings obtained with the two slides were compared. Figure 1 gives a schematic view of rosette formation and a photograph of one rosette in a stage 2 patient taken in the field. Schematic approach of the CD19 pan B rosette technique. (a) (left panel): B lymphocyte expressing CD19 antigen and CD19 negative mononuclear cells in cerebrospinal fluid. (b) (right panel): addition of anti-CD19 antibody beads (Dynabeads® CD19 pan B) triggers the formation of a rosette due to the fixation of at least four beads to the CD19 sites of the B lymphocyte. (c) The photograph at the right bottom corner was from a stage 2 patient examined in the field, using an Optio E40 Pentax camera positioned on the eye piece of a microscope. It shows one B lymphocyte conglomerated with Dynabeads® CD19 pan B and three free beads in the patient’s CSF examined on a KOVA® slide. Cerebrospinal fluid cell and rosette counts obtained with Nageotte versus KOVA® slides were compared by simple regression analysis in patients from groups I and II. Comparisons between HAT stages used non-parametric tests based on median values because of the limited size of patient categories (stages 1 and 2, and intermediate stage). However, for the sake of clarity, mean and SEM values are also given in Table 1. The Kruskal–Wallis univariate rank test (three-tailed) was followed by Mann–Whitney U-tests (two-tailed) as post-hoc tests. Follow-up was analysed using Wilcoxon signed rank test, considering each patient as his own control. The minimum level of significance was set at 0.05 for all analyses. During the initial diagnosis procedure in groups I and II, total cell counts obtained with Nageotte vs. KOVA® slides were highly correlated (n = 49; r = 0.999; P < 0.001). A similar relationship was found for rosette counts (r = 0.996; P < 0.001). Readings were therefore similar with the two slides. Consequently for simplification, only KOVA slide values are shown in the Results section. Figure 2 (left panel) gives individual values obtained in the 71 patients of the three groups during the diagnostic procedure. Individual patient’s codes are presented for those with CSF cell counts >5/μl. The 16 stage 2 patients were treated with eflornithine (10 cases) or melarsoprol (six cases). The 13 intermediate stage patients and 42 stage 1 patients received pentamidine. Total cell and rosette counts for each HAT stage are given in Table 1 and Figure 2 (left panel). Kruskal–Wallis analysis indicated a stage effect on cell counts (P < 0.001). Using post-hoc Mann–Whitney tests, the 16 stage 2 patients exhibited higher cell and rosette counts than the 42 stage 1 (P < 0.001) and the 13 intermediate stage patients (P < 0.001). Cell counts in intermediate stage patients were significantly different from stage 1 patients (P < 0.001). Similarly, a stage effect (Kruskal–Wallis test) was observed on rosette counts (P < 0.001). Rosette count values differed in stage 2 compared to stage 1 (16 stage 2 vs. 42 stage 1 patients, P < 0.001) and intermediate stage patients (n = 13, P < 0.001). There was no difference in rosette counts between stage 1 and intermediate stage patients. Thirty seven stage 1 patients demonstrated no B lymphocytes in their CSF and one B cell/μl was counted in five patients (S60, S105, S141, S146, S154), patient S105 having a haemorrhagic CSF sample. Conversely, stage 2 patients showed rosette counts of at least 2/μl, except for S97 (24 cells/μl, no rosette) and S158 (20 cells/μl, one rosette). Nine intermediate stage patients counted no B lymphocyte in the CSF. However, three patients (S142, S161, S164) exhibited one CD19 positive cell/μl, while S157 exhibited two B lymphocytes/μl. Of the 50 patients of groups I and II, 40 (80%) were followed-up 6 months after treatment, 38 (76%) participating in the 12-month follow-up. From the 23 patients of group I, 16 (70%) attended the 18-month follow-up, and 10 (44%) were present at the 24-month follow-up. Figure 2 gives individual time courses in patients from both groups with patient’s codes being indicated when CSF cell counts scored higher than 5/μl or rosette counts higher than 1/μl during the diagnosis evaluation. There was a difference (Wilcoxon signed rank test) in total cell counts between the initial diagnosis values and the counts performed after a follow-up of 6 months (P < 0.001), 12 months (P < 0.01), 18 months (P < 0.05), and 24 months (P < 0.05). Such a difference indicated that there was a clear improvement of patients’ condition in the follow-up examinations (Figure 2; Table 2). However, no significant differences were observed between post-treatment sessions indicating that the health improvement due to treatment was obtained within the first 6 months. Similar findings were drawn considering each diagnostic stage category. In stage 1 patients, cell counts were ameliorated in the follow-up procedure compared to values at diagnosis (6-month follow-up, n = 15, P < 0.05; 12-month follow-up, n = 8, P < 0.05; 18-month follow-up, n = 9, tendency at P = 0.06). Follow-up examinations did not differ from each other. For the intermediate stage, the size of the patient sample only allowed comparisons between diagnosis and 6-month (n = 7, P < 0.05) and 12-month (n = 8, P < 0.05) follow-up values, also showing an improvement of health condition after treatment. In stage 2 patients, cell counts differed between diagnosis and 6-month (n = 11, P < 0.01) and 12-month (n = 11, P < 0.05) follow-up tests. Similarly, the comparison of rosette count values in the diagnosis and follow-up procedures showed a difference at the 6-month (n = 11, P < 0.01) and 12-month (n = 9, P < 0.05) treatment follow-up evaluations, indicating normalization of CSF composition, with a decrease or disappearance of B lymphocytes. Immune cells and trypanosomes are known to cross the blood–brain barrier in experimental African trypanosomiasis (Masocha et al. 2004). However, the nature of immune cells penetrating the CNS has been rarely analysed and has not yet been investigated in field conditions. The present study was conducted to determine whether the type of CSF cells in HAT, especially B lymphocytes, could be examined in African village conditions. One of the goals of the study was to rule out the intermediate stage, considering that the presence of two and more rosettes would signify the alteration of neurological functions (stage 2), thus improving diagnosis decision and treatment adequacy. The choice of the Dynabeads® CD19 pan B rosetting technique to identify CSF B lymphocytes was based on the pioneering work by Greenwood et al. (1976) showing that B lymphocytes presenting immunoglobulin M (IgM) type antibodies represent up to 30% of CSF cells in CNS inflammation. In stage 2 HAT, they reported increased immunoglobulin-bearing lymphocytes in the CSF. They identified T lymphocytes with the sheep red cell rosetting test and B lymphocytes through the presence of surface immunoglobulin using anti-whole immunoglobulin conjugate and stimulation by pokeweed mitogen, a B cell mitogen. Recently, Boda et al. (2009) used flow cytometry and found 5.9% of CD19 B cells in the CSF of eight intermediate stage patients and 51.1% in 13 stage 2 patients. Inflammatory CNS alterations, such as in multiple sclerosis, are known to trigger antibody responses (Cepok et al. 2005). First evidenced by Mattern (1968), an increased CSF IgM concentration is considered to be a major marker of stage 2 HAT (Lejon & Büscher 2005). However, CSF IgM detection in the field is still under investigation (Lejon et al. 2008). Furthermore, only a small proportion of IgM is directed to trypanosome antigens, and antibodies directed to CNS components are present in HAT. In murine experimental allergic encephalomyelitis, injection of anti-galactocerebrosides exacerbates CNS inflammation and clinical severity (Morris-Downes et al. 2002), suggesting that autoantibody and/or cytokine production by CSF B lymphocytes might be involved in the pathophysiological events observed in HAT. Detection and quantification of the large CD19 B rosettes using a small amount of CSF was easily performed. An increase in CD19-expressing cells was found in HAT patients, proportionally to the evolutionary stage of the disease. Stage 1 patients exhibited either zero (37 out of 42 patients) or one rosette/μl (patients S60, S105, S141, S146 and S154). Therefore, values observed in stage 1 patients are comparable to flow cytometry data in healthy volunteers showing that CD19 B lymphocytes represent 0.8% of total CSF cells (Svenningsson et al. 1995). Patients with CSF white cell counts conventionally corresponding to stage 2 exhibited at least 2 B lymphocytes/μl. Nine of the 13 intermediate stage patients did not have any CD19 cell in their CSF. However, none of these patients did belong to the uncertain evolution group during post-treatment follow-up. A threshold of 1 rosette/μl may be considered as the limit of pathology for the CNS, since in patients with favourable evolution B cell counts remained at or tended to resume values of zero to one CD19 positive cell/μl, while at least two B lymphocytes were detected in uncertain evolution and relapse at stage 2 patients. Two stage 2 patients showed either one CD19 positive cell/μl of CSF (S97) or none (S158). Both patients exhibited white cell counts in the vicinity of 20 cells/μl. In such borderline cases, the question can be raised as to whether stage 2 medications should be applied. Conversely, one intermediate stage patient revealed 2 B lymphocytes and 7 white blood cells/μl of CSF (S157). As in several previous reports, clinical data obtained from our 71 patients confirmed that neurological symptoms and signs lack specificity and should not solely be used for staging (Chappuis et al. 2005). Neurological signs were observed in any disease stage, but may have been absent in some stage 2 patients. Accordingly, the absence or presence of symptoms did not correlate with rosette counts. In conclusion, the use of the rosette technique during diagnosis demonstrated that stage 1 patients exhibited zero (37 out of 42 patients) to 1 rosette/μl (five patients) of CSF. In contrast, B lymphocytes were present in the CSF of the large majority of stage 2 patients (≥2 rosettes/μl), indicating ongoing CNS infection. In our patient sample, patients in the intermediate stage category did not show any abnormal rosette count and experienced favourable evolution during post-treatment follow-up. Only one patient (S157) did show 2 rosettes/μl of CSF. However, this patient had not yet undergone the follow-up procedure. During the follow-up procedure, B lymphocytes were not observed in patients with favourable evolution but were present in patients with uncertain evolution and/or relapse at stage 2. We are inclined to believe that CSF B cell rosettes, being easily detected, can limit uncertainty in separating stages 1 and 2 HAT, especially in patients with intermediate stage. We acknowledge the patients’ assiduity in attending investigations every 6 months which obliged them to live far from their families and home villages repeatedly. We are also indebted to the national authorities of the Republic of Congo for their assistance, encouragement and help in setting up the investigations. We recognize the indispensable logistical support from Mr Raymond Mboulou, Member of Parliament from the District of Mpouya, Minister in the government of the Republic of Congo and President of Omitry, a Congo non-governmental organisation. We also thank Mr Charles Zacharie Bowao, Minister of Cooperation, for improving the comfort of our patients. We especially thank for their continuous support and/or assistance Prof. Henri Joseph Parra and Dr Etienne Mokondji Mobé from the National Public Health Institute, Dr Bébène Bandzouzi from the Brazzaville University Hospital (Neurology ward). [Correction made here after initial online publication]. Jean-Claude Bambi and Dieudonné Mouanda from the National Control Programme for Human African Trypanosomiasis gave valuable technical and nursing assistance. Mr François Obanda and Mr Jules-César Botokou Eboko, Attachés to the Presidency of the Republic of Congo, took care of administrative and logistical matters throughout the ongoing investigation. Prof. Manny W. Radomski of the University of Toronto (Canada) gave constructive comments on the manuscript. The project was funded by WHO/TDR grant # A50468 and partner universities of Bordeaux, Limoges and Lyon.