journal of Tropical and Geographical Neurology 1992; 2: 102- 108
ORIGINAL ARTICLE
A geographical cluster of konzo in Tanzania
W P Howlett OTM&H FRCP I1 G Brubaker M02, N Mlingi BSc Msc3 and H Rosling MO Ph04 'Kilimanjaro Christian Medical Center, Moshi, Tanzania; 2Shirati Hospital, Musoma, 3Tanzania Food and Nutrition Center, Dar-es-Salaam, Tanzania; and 4the International Child Health Unit, University Hospital, Uppsala, Sweden Keywords: Konzo, epidemic, cassava, cyanide, motoneuron disease, Tanzania
Summary cas ont ete affectes pendant une flambee epidemique a la fin d'une secheresse en 1985. La distribution Konzo, an upper motor neuron disease, was diag geographique et temporelle est compatible avec une nosed in 116 subjects in a community-based survey etiologie alimentaire proposee ayant comme facteur of 28,500 inhabitants in the Tarime District in principall'exposition au cyanure. Cette ingestion Tanzania. All cases had a uniform clinical picture de cyanure provient des racines ameres de manioc with isolated non-progressive paraparesis of abrupt insuffisamment preparees. D'autres facteurs d'ordre onset identical to the findings in earlier studies of toxico-nutritionel peuvent etre impliques mais konzo. The fishing population along Lake Victoria l'epargnement de la population de pecheurs parle was not affected but farming villages only 5 km from contre une etiologie infectieuse, comme par le the shore had a maximum prevalence of 14 per 1000 HTLV-l. inhabitants. New cases of konzo have occurred annually since 1979 but 62 % were affected during Resumen an epidemic at the end of a drought in 1985. The geographical and temporal distribution is Konzo es una enfermedad de la motoneurona compatible with the proposed dietary aetiology of superior descrita en Africa. En el presente estudio cyanide exposure from insufficiently processed se diagnosticaron 116 pacientes con konzo en un cassava roots as the main factor. Other toxico estudio de prevalencia en una comunidad de 28,500 nutritional factors may also be involved. An habitantes en el Distrito Tarime de Tanzania. aetiological role of HT LV-I has earlier been excluded Todos los pacientes presentaban un cuadro clinico and the exclusion of the fishing population makes similar caracterizado por la presencia de una an involvement of another type of infectious agent paraparesia no progresiva de comienzo abrupto, sin unlikely. otro compromiso neurologico, similar a la descrita en estudios previos del konzo. No se encontraron Resume casos en las poblaciones de pescadores de las orillas del Lago Victoria. Sin embargo, en las Konzo, une affection du neurone moteur superieur, comunidades agricolas situadas tan solo a 5 km del a ete diagnostique chez 116 malades parmi les lago las cifras de prevalencia a1canzaron valores 28,500 habitants du district de Tarime en Tanzanie. hasta de 14 por 1000 habitantes. Nuevos casos Tous le cas s'etaient presentes une symptomatologie de konzo continuan presentandose anualmente uniforme caracterisee par la survenue brutale d'une desde 1979, pero el 62% de los casos ocurrieron paraparesie spasmodique isolee et non-progressive, en forma epidemica hacia el final de la sequia identique a celle rapportee dans des etudes de 1985. La distribucion geografica y temporal de anterieures du konzo. La population de pecheurs au los casos es compatible con la hipotesis de que bords du Lac Victoria n'etait pas touchee par le el konzo se debe a una posible causa dietetica konzo mais dans les villages agricoles, a une distance cuyo factor etiologico mas import ante es la de 5 km on trouve une prevalence maximale de 14 exposicion al cianuro por la dieta exclusiva de sur 1000 habitants. Des nouveaux cas de konzo sont tuberculos de la yuca (mandioca 0 casava) procesada survenus annuellement depuis 1979 mais 62 % des en forma inadecuada. Otros facto res toxicos nutricionales pueden tambien intervenir. Correspondence to: H Rosling MD. International Child Health Anteriormente se demostro que el retrovirus HTLV-J Un it, University Hospital, Entrance 1 I , 5-751- 85 Uppsala, no juega un papel etiol6gico en casos de konzo. Sweden. Por otra parte, la ausencia de la enfermedad Howlen et al. A geographical cluster of konzo in Tanzania 103 en las poblaciones pesqueras estarfa en contra de relatively homogeneous agro-ecological charac una etiologfa infecciosa. teristics with a landscape varying between 1100-1200 m above sea level in the fishing villages along the shore of Lake Victoria. Some hills reach Introduction 200 m higher in the inland villages where the Recent reports from three African countries strongly population depends on farming in spite of a indicate that konzo is a distinct form of upper motor relatively low soil fertility. The inhabitants live in neuron disease 1-3. Konzo is named after the local large villages created during the villagization designation in the first reported epidemic4 . The programme in the mid-70s, and during the last disease is characterized by an abrupt onset of decades cassava has become both their main staple isolated and permanent but not progressive spastic and their cash crop. They mainly grow bitter paraparesis. The paralysis is always symmetrical, but cassava varieties that were introduced in 1979-1980 may vary in severity from mild gait difficulties to because of high yields and drought tolerance. Before a severe paralysis of the legs, the trunk, and consumption the cyanogen substances are normally occasionally the arms. The most severely affected removed from these roots by processing, as subjects may in addition have dysarthria and visual described elsewhere5. disturbances5. We surveyed the 15 northern villages in the Lathyrism is the disease most similar to konzo, but district (Figure 1) in April 1989. The number of there is no evidence for the consumption of inhabitants in each village was obtained from local lathyrus sativus in any of the konzo-affected census records and rounded to the nearest hundred. populations6 Konzo has been attributed to several With informed consent and assistance from village weeks of almost exclusive consumption of cassava leaders and village health workers, all subjects with roots that because of short cuts in processing have gait difficulties were identified in each village and high remaining amounts of cyanogen substances. It examined in their homes by a physician (WPH, HR). has been postulated that the resulting dietary Subjects with non-neurological locomotor disorders cyanide exposure in combination with low intake such as arthritis, skeletal deformities, and rickets of sulphur needed for detoxification selectively were excluded. With a pretested questionnaire each damages the upper motor neurons through some subject was asked in Lou through an interpreter unknown mechanism'- about type of onset, place of residence at onset, the A clinical study of konzo during an outbreak in course of the disease, and possible occurrence of the Tarime District of Tanzania in 1985 suggested subsequent attacks. The year and month of onset that the epidemic was related to prolonged drought. of paralysis was determined by the use of a local This study in combination with dietary and event calendar. Women in a fertile age were also laboratory results supported a causative role of asked about how the onset related to pregnancies cyanide in konzo 5 . An infectious cause has also and breast feeding. Abnormalities of gait and been proposedB, but konzo differs clinically from running, number of walking sticks in regular use, the progressive form of spastic paraparesis induced and speech abnormalities were observed. The by HT LV-I infection. The 39 konzo cases studied clinical and neurological examination was mostly earlier in Tarime District were all seronegative to carried out in the sitting position. HTL V_1 5 as were all konzo cases earlier studied in The criteria for a diagnosis of konzo l were as other areas2.3.9 . follows : (1) the presence of a visible symmetrical We have surveyed the population in the northern spastic abnormality of gait and/or running; (2) a part of Tarime District for locomotor disabilities history of distinct onset in time in a previously four years after the konzo epidemic to determine healthy person; (3) non-progression; and (4) the if the geographical and temporal distribution of presence of bilaterally exaggerated tendon reflexes konzo fitted the proposed toxico-nutritional in the legs. A clinical diagnosis was also made for aetiology or if the occurrence of the disease persons suffering from other neurological diseases suggested the involvement of an infectious agent. causing gait difficulties, and when necessary they were referred to Shirati Hospital for treatment. Monthly rainfall data from 1960 to 1990 was Methods obtained from Shirati Hospital. Tarime district is situated in northern Tanzania and borders Kenya. The study area in the dry northern Results part is on the shore of Lake Victoria. It is inhabited by the Luo tribe and has, compared to other parts In the 15 villages surveyed we identified 208 of rural Tanzania, a fairly dense population with subjects with neurological locomotor disabilities, of about 70 inhabitants per km2. The study area has whom 116 fulfilled the criteria for konzo (Table 1). 104 Journal of Tropical and Geographical Neurology Volume 2 Number 3 1992
..r€/f/.!: <1
~ Affected area
~ Mainroad Skm ~ Intervillage border
Figtlre I. The geographical distribution of konzo. The ntlmber of konzo cases and inhabitants, and in parenthesis the prevalence per 1000 inhabitants, is given for each of the 15 studied villages
Five subjects with konzo who lived just outside the were excluded. Excluded also were two earlier study area at the time of the survey were included verified patients with konzo in inland villages who as they had lived in the study area at the onset. Two had died before the survey and seven persons konzo patients who were temporarily staying with suspected of having konzo in inland villages who relatives in shore villages, but who had lived in had either moved from the study area or who were villages east of the study area at the time of onset, unavailable for examination for other reasons. A further six subjects in inland villages presented a Table I. Number of persons with neurological locomotor history of abrupt onset of mild spastic gait problems disabilities in the shore and inland villages typical for konzo; they had bilaterally exaggerated leg reflexes but were excluded because they had no Shore villages Inland villages visible disturbance of gait or running at the time of No. of villages 5 10 examination. The semi-urban community surround Population 9400 19,100 ing Shirati hospital was not surveyed since it was Konzo 11 105 well known that none were affected by konzo in this Polio sequelae 29 30 population that consisted of mainly hospital staff. Hemiparesis 3 6 Paraparesis 3 7 A total of 91 subjects with gait difficulties were Quadriplegia 3 4 found to have other neurological disorders that are Polyneuritis 2 I listed in Table 1. The sequelae of poliomyelitis was Others 0 4 diagnosed in 59, of whom 24 had right and 21 had Total 51 157 left lower limb paralysis; the others had combined Howlett el al. A geographical cluster of konzo in Tanzania 105 forms of paralysis. No case of polio had an onset breastfeeding, and one had the onset two years after after 1983, when vaccination was started. Of those delivery. with hemiparesis three had an infantile onset. The In 14 % of the 116 konzo cases we observed only differential diagnosis between konzo and other a spastic abnormality on running. Fifty-three per disorders was clear in all subjects seen in the shore cent had a spastic gait but did not require sticks; villages. Two of the three patients with paraparesis 19% used one stick to walk, 9% used two sticks; in the shore villages were diagnosed as having and a further 5% were unable to stand. All had tuberculosis with a typical history, gibbus and bilaterally exaggerated knee reflexes. Bilaterally corresponding sensory level on the trunk. The third exaggerated ankle reflexes were found in 92 %, one had spastic diplegia present since birth. Two of the had a unilaterally exaggerated ankle reflex; and the cases with quadriplegia in the shore villages were remaining 7% had normal ankle reflexes. Bilateral congenital and diagnosed as cerebral palsy, and the non-sustainable ankle clonus was noted in 17 %, third had resulted from a severe acute cerebral sustainable clonus in 67 %, unilateral non infection in infancy. A case of sensory neuropathy sustainable in two and non-sustainable on one side of five month's duration was seen in a shore village and sustainable clonus on the other side was found in a 65-year-old woman, and a relapsing post in three patients. Ankle clonus was absent in 9% and inflammatory polyneuritis was diagnosed in a could not be adequate tested in three cases. Plantar 12-year-old girl. reflexes were bilaterally extensor in 72 %, uni In inland villages the differential diagnosis of laterally in one patient, and flexor in 24 % and were konzo was difficult only in four cases. A 28-year not tested in two patients. With only one exception old man with spastic paraparesis, weakness in the all of the 32% of the patients who used one or two arms, and dysarthria since 1986 was not diagnosed sticks or were unable to stand had bilaterally as having konzo due to slow onset and steady exaggerated ankle reflexes and sustained ankle progression. A 10-year-old boy had permanent clonus, and bilateral extensor plantar reflexes were isolated spastic paraparesis but did not fulfil the observed in all but three. criteria for konzo because of a very slow onset Minor difficulties in vision were reported by six during which he lived in Dar-es-Salaam. A 47-year patients and eight patients had spastic speech old woman was unable to stand because she had abnormalities and of these four and six, respec severe spastic paraparesis. She had an abrupt onset tively, were unable to walk without support. in February 1989 but was not diagnosed as having Kyphoscoliosis was seen as a sign of trunk paralYSis konzo as she gave a clear history of successive in severely affected cases, but no other spine progression during the last months. A 4-year-old abnormalities were found. Minor leg ulcers were boy with isolated spastic paraparesis did not fulfil noted in seven cases but none had abnormal sensory konzo criteria since he was born with the disorder functions. None had oedema and only one had a in May 1985. Interestingly, his mother had a first small nodular goitre. attack of konzo in 1983 and had an exacerbation The geographical distribution given in Figure 1 one month before his birth. shows that all inland villages were affected by konzo Of the 116 konzo patients 67 % stated that the with a prevalence of 14 per 1000 inhabitants in the onset occurred abruptly within one day, mostly most affected village. No case of konzo was found without any other symptoms. The remaining 33% in the population living within three km of Lake gave a history of subacute onset extending over Victoria. The families of all the 11 patients with several days. Five patients stated that they had konzo who belonged to the shore villages (Table 1) improved slightly some months after the onset and were farming households and stayed more than the other 96% stated that the paralysis had three km from the shore in hilly parts adjacent to remained unchanged since the onset several years inland villages, as marked in Figure 1. earlier. Twelve patients gave a clear description of The age at onset versus sex distribution given in a subsequent attack that left them with an increased Figure 2 reveals that none had been crippled by disability. One woman had suffered two such konzo during the first two years of life. Below the attacks. Konzo in one first or second degree relative age of 20 years konzo was more common in men but was reported by 29% of the patients; an additional at higher ages slightly more women than men had 12 % had two or more relatives affected; and in one been affected. family four of five children had been disabled by Figure 3 shows that the first incidence of konzo konzo. In 11 of 13 adult women information was occurred in 1979 and was followed by a slowly obtained on the relation between the onset and increased annual incidence until the epidemic delivery of a baby. Only two had the onset during outbreak in 1985, during which 62% of the pregnancy; five during the first four months of recorded cases occurred. The start of the epidemic breast feeding, three later during the first year of in February 1985 followed a drought in 1984 that 106 Journal of Tropical and Geographical Neurology Volume 2 Number 3 1992
'" 10 continued into 1985. Figure 4 shows that the c;;'" epidemic occurred in the last dry month and the E 8 ~ first three months of the rainy season. During the ..... 6 other years the incidence of konzo was also slightly 0.... B 4 higher during the same season. E 2 ::I Z 0 0 10 20 30 40 50 Discussion 14 The use of paramedical staff for identification of '" neurological diseases in epidemiological surveys is C;;'" 10 E customary in developing countries 10. However, as ..... 8 ....0 diseases like konzo result in an easily observed gait 6 abnormality of long duration, we successfully oD'" E utilized village leaders in our identification process. ::I Z In inland villages some subjects with other neurological locomotor diseases may have been missed as village leaders perceived the study as a 50 search for konzo, a disease they could distinguish Age at onset from other locomotor disabilities. The higher Figure 2. Distribution ofsex and age at onset oftbe 116 konzo prevalence of non-konzo disabilities found in shore cases villages may also partly be explained by an additional intensive house-to-house search that was made along the shore to verify the complete absence of konzo in this area. ~ The clinical pattern of konzo was very similar l 8 a 60 1200 to that reported from other affected areas • ,11 . " To distinguish konzo clinically from other causes '"k " 1000 ~ of paralysis was easy. The proportion of the ""~ ~ 40 800 ~ various degrees of disability was almost identical "u S @lj to that found in Zaire, as was the occurrence of 0 problems of speech and vision in some of the " .5 ~ 20 I . '"0 400 ::l most severe cases The uniform clinical picture ] !! of abrupt onsets, isolated and symmetrical E 200 .~ ~ spastic paraparesis without any progress ions but Z " 0 occasional occurrence of subsequent aggravating 74 75 76 7178 79 80 81 82 83 84 85 86 87 88 89 attacks, supports the view that konzo is a distinct Year disease entity almost exclusively affecting the Figure 3. Annual distribution of new konzo cases (bars) and function of the upper motor neurons. The clinical annual rainfall (dots) picture of konzo clearly differs from that of HTL V-I associated myelopathy, a disease with slow onset and progressive course. However, subclinical ~ 30 forms of konzo that lack a clear spastic abnormality -5 c of running are obviously arbitrarily distinguished 0 ...e from mild forms of the disease. If the entire g, ~ population could have been examined we might 200 -5 '" § also have identified more mild cases of konzo, but .."'"