Pathogenetic mechanisms in dengue haemorrhagic fever: Report of an international collaborative study*

Memoranda Pathogenetic mechanisms in dengue haemorrhagic fever: Report of an international collaborative study * In a study of 55 persons with dengue haemorrhagic fever-36 of whom showed the dengue shock syndrome-clinical, haematological, virological, and serological changes were correlated with serial measurements of complement components and immunopathological studies. Viruses dengue-l or dengue-2 were isolatedfrom the sera of9 patients. Serological responses indicative of secondary dengue virus infections were observed in 53 patients; 2 (infants) hadprimary infections. During the acute phase of the disease, dengue antibody titres rose logarithmically. Marked depression of complement components, especially C3, was observed. Activation of both the classical and alternative complement pathways was demonstrated, with depression of both C4 and C3 proactivator levels in most instances, although in some cases it appeared that one mechanism was involved to a greater extent than the other. The level ofdepression ofC3 was correlated with the severity of the disease. Relatively stable transferrin levels indicated that depletion of complement proteins was not primarily due to extravasation. Fibrinogen levels were depressed and fibrinogen split products were found in the plasma. The accumulated data provide further evidence of the central role that activated complement components play in the pathogenesis of dengue haemorrhagic fever. The pathophysiological processes that occur in dengue haemorrhagic fever and that distinguish it from unmodified dengue fever are increased vascular permeability, decreased plasma volume, hypoten- sion, thrombocytopenia, and a haemorrhagic dia- thesis. Although the mechanism(s) responsible for the distinctive pathophysiology of dengue haemor- rhagic fever are incompletely understood, the rapid onset of these short-lasting and often rapidly rever- sible pathophysiological processes, and the fact that histopathological vascular lesions have not been found, suggest that pharmacologically active media- tors play an important role in the production of this syndrome. The association of dengue haemorrhagic fever with secondary, heterologous dengue virus infections (Halstead et al., 1967; Winter et al., 1968) suggested the possibility that immunopathological processes are the determining factor in whether infection with * The names of the participating institutions are listed on page 133. The study received support from the World Health Organization, the USA-Japan Cooperative Medical Science Program, and the US Army Medical Research and Development Command. a dengue virus results in unmodified benign dengue fever or in dengue haemorrhagic fever. Support for this concept is found in the observation that anam- nestic antibody responses occur in dengue haemor- rhagic fever, resulting in a high concentration of antidengue IgG antibody in the blood early in the course of the disease (Russell et al., 1969). The occur- rence of circulating antigen-antibody complexes in dengue haemorrhagic fever has been postulated (Russell, 1970, 1971) on the basis that both antibody and infectious virus are present in the blood early in the course of the disease. Studies of patients with dengue shock syndrome- the most severe form of dengue haemorrhagic fever- demonstrated that serum levels of the third com- ponent of complement (C3) are markedly reduced, suggesting that complement consumption occurs in vivo (Russell et al., 1969). The basic reason for the depression of C3 is not known, but it seemed logical to postulate that the low levels may result from the activation of the complement system by immune complexes. Since activation of the complement sys- tem may lead to the formation of anaphylotoxins (C3a, C5a) a hypothesis was framed (Russell, 1970, 2988 -117- MEMORANDA 1971), according to which C3a is the mediator that causes the changes in vascular permeability and hence is responsible for the shock. Thrombocytopenia and moderate depression of several clotting factors occur in dengue haemor- rhagic fever, but at present there is no adequate single explanation for the haemorrhagic diathesis. Although lowered plasma fibrinogen levels during dengue haemorrhagic fever have been interpreted as evidence for disseminated intravascular coagulation, the relative importance of the latter as a major pathogenetic mechanism is controversial. Recent advances in knowledge of the biochemical pathways through which the complement system is activated, and of the physiological activity of acti- vated complement components, offered an oppor- tunity to apply basic biochemical knowledge to the understanding of a disease process of major public health importance. The development of sensitive immunochemical techniques for the quantification of complement proteins made such a study feasible. This collaborative study was carried out with the primary purpose of observing the changes that occur in the complement system in persons with virologi- cally proved dengue haemorrhagic fever and relat- ing changes in the serum levels of complement com- ponents to antidengue antibody levels, haematologi- cal changes, and the clinical course of the disease. The study also aimed to demonstrate the presence of immune complexes in the blood and tissues, and to find out whether changes in the factors concerned with the clotting mechanism might be correlated with the other observations. METHODS Study population Persons with dengue haemorrhagic fever admitted to the paediatric wards of the 4 collaborating hos- pitals in Bangkok, Thailand, between 1 July and 15 September 1971 comprised the study population. During those 3 months epidemic dengue haemor- rhagic fever was prevalent in Bangkok. Concurrent epidemiological studies indicated that over 500 cases requiring hospitalization occurred in the Greater Bangkok area and that at least 3 dengue serotypes- dengue-1, dengue-2, and dengue-3-were present (F. H. Top, unpublished observations). Admission to the study was based on the initial clinical diagnosis of dengue haemorrhagic fever, and patients with shock or impending shock were selected for study in preference to those with a milder form of the disease. Detailed clinical records were kept on each patient. Blood specimens were obtained daily during the acute phase of the disease for haemo- globin, haematocrit, leucocyte and platelet counts, fibrinogen and fibrinogen split products assay, virus isolation, virus serology, complement component determinations, and immunochemical studies, in addition to those tests essential to the management of the patient. Therapy in most cases included fluid and electrolyte replacement and maintenance of the plasma volume with serum albumin or fresh plasma. Whole-blood transfusions were needed in some cases. A standard chart for pertinent signs, symp- toms, clinical laboratory findings, and treatment was maintained at each institution. At the conclu- sion of the study the charts were reviewed and the clinical course of the disease in each patient was independently classified by tho clinicians in charge of the case. The following criteria (Nimmannitya et al., 1969) were used to grade the study population into 4 groups on the basis of the severity of the disease: Grade I. Fever accompanied by nonspecific con- stitutional symptoms, with a positive tourniquet test as the only haemorrhagic manifestation. Grade IL. Fever and skin haemorrhage or other bleeding, such as epistaxis or gum bleeding. Grade III. Circulatory failure manifested by rapid, weak pulse with narrowing of the pulse pressute (<20 mm Hg) or hypotension. Grade IV. Moribund patients with undetectable blood pressure and pulse. Serum specimens for comparison and control of the complement determinations were also obtained from 30 "normal" Thai children and from 13 patients admitted to the Women's and Children's Hospital with acute infectious diseases other than dengue, including measles, mumps, and several bac- terial infections. Virology and serology Attempts were made to isolate the virus from the initial serum samples obtained from all the patients, by the direct and delayed plaque technique in LLC-MK2 cells (Yuill et al., 1968). The serum was inoculated freshly or after storage at -70°C. Isolates were identified by the plaque reduction neutraliza- tion test with reference monkey antiserum to proto- type strains (Russell & Nisalak, 1967). Haemagglu- tination-inhibition (HI) tests were carried out as described by Clarke & Casals (1958) with a modifica- 118 PATHOGENESIS OF DENGUE HAEMORRHAGIC FEVER 119 tion to the microtitre. Antigens prepared from dengue-l (Hawaii), dengue-2 (New Guinea C), den- gue-3 (H-87), dengue-4 (H-241), Japanese encephal- itis (Nakayama), and Chikungunya (Ross) viruses were used. Acetone-extracted sera were tested against 8 units of antigen. All the sera from each individual patient were tested simultaneously. Complement studies For the immunochemical quantification of com- plement proteins, the single radial immunodiffusion method of Mancini et al. (1965) was used. Mono- specific antiserum to human complement proteins was incorporated at a previously determined, opti- mum concentration into 1.5% agarose gel contain- ing isotonic phosphate buffer, pH 8.0, and 0.O1M EDTA. Seven ,l of 3 different test serum dilutions were placed in 3-mm wells and the diameter of the precipitin rings was measured after 24 hours at room temperature. Each immunoplate received 5 dif- ferent dilutions of the standard reference serum. The absolute amount of a complement protein in a given test serum was determined by means of a graph relat- ing the square of the radius to the concentration of the standard serum. The concentration of various complement proteins in the standard serum had pre- viously been quantified 1 with highly purified proteins as reference substances. Methods have been described for isolating Clq (Calcott & Miiler-Eberhard, 1972), Cls (Haines & Lepow, 1964), C3 (Nilsson & Miiller- Eberhard, 1965), C4 (Miller-Eberhard & Biro, 1963), CS (Nilsson & Miiller-Eberhard, 1965), C6 (Arroyave & Miiller-Eberhard, 1971), C8 (Manni & Muller- Eberhard, 1969), C9 (Hadding & Miiller-Eberhard, 1969), and C3 proactivator (C3PA) (Gotze & Muller- Eberhard, 1971). Antisera were made in rabbits or goats. Transferrin was assayed with reagents made available by Dr Ursula Miiler-Eberhard. Haemolytic assays were performed by microtitre techniques with sheep erythrocytes (E) and rabbit antibody to E (A). CH50 was determined with EA according to standard methods; C2H50 was estimated with EACI,4 as assay cells and EDTA guineapig serum as the reagent (Cooper et al., 1970); C6 was determined with EA and with C6-deficient rabbit serum (Rother et al., 1966). For the detection of preformed conversion pro- ducts of C3 and C3PA, EDTA plasma was exam- ined by immunoelectrophoresis. The functional inte- grity of the alternative pathway of complement 1 At the Scripps Clinic & Research Foundation, La Jolla, Calif., USA. activation was qualitatively evaluated by the inulin test (Gotze & Miiller-Eberhard, 1971). Clq studies Clq prepared as previously described (Calcott & Miiler-Eberhard, 1972) was used in an attempt to detect materials in the patients' sera capable of activating the C sequence. As a screen a precipitin test was employed in which 20 pl of the patients' undiluted serum overlaid 20,ul of Clq (1 mg/ml) in capillary tubes measuring 1.6-1.8 x 100 mm. The tubes were kept at room temperature for 4 days and observed daily for the development of precipitates. Precipitates filling the angle of the lower fluid menis- cus were graded +; those covering the entire menis- cus surface + +; and larger amounts ++ +. This procedure produced traces of precipitate (±) in half of the control sera from healthy Thai children, whereas the remainder of the control sera were negative except for a rare + reaction. All available sera from each patient were tested in order to observe changes in Clq reactive material throughout the course of the disease. Clotting studies The extent to which the clotting system was involved in the dengue shock syndrome was evalu- ated by quantifying the levels of clottable fibrinogen and split products of fibrinogen plus fibrin in the plasma of all patients throughout the course of the disease. A radial diffusion method utilizing antiserum to fibrinogen was employed (Edgington, 1972). Immunofluorescence studies Direct immunofluorescence studies were performed as described by Wilson & Dixon (1970), with the following fluorescein-labelled antisera: rabbit anti- human immunoglobulins, rabbit antihuman C3, rab- bit antihuman fibrinogen, rabbit antihuman albumin, serum from patients convalescing from dengue fever (convalescent serum), and mouse antidengue-2. RESULTS1 The results for 55 patients with virological or serological evidence of dengue are included in this 1 The results of the multiple laboratory determinations made on the patients in this study are much too extensive to report in detail here. This report summarizes the data obtained and individual data are given only for selected cases. Single photocopies of the complete tabulated results may be obtained by professionally interested persons on request to: Chief, Immunology, World Health Organization, 1211 Geneva 27, Switzerland. MEMORANDA Table 1. Summary of clinical signs and laboratory findings in 55 dengue patients Severity of disease Findings Grade I (5)a Grade 11 (14) Grade III (23) Grade IV (13) fever 100 100 100 100 hepatomegaly (2-5 cm) 100 (4/4)b 91 (10/11) 100 (20/20) 100 (13/13) positive tourniquet test 50 (1/2) 92 (11/12) 84 (16/19) 62 (5/8) petechiae 0 100 (12/12) 52 (12/23) 69 (9/13) epistaxis 0 0 17 (4/23) 8 (1/13) haematemesis/melena 0 0 13 (3/23) 69 (9/13) haemoconcentration C 60 (3/5) 71 (10/14) 91 (21/23) 69 (9/23) platelet counts: 100-150 x 103 per mm3 20 (1/5) 15 (2/13) 0 8 (1/13) 50-100 x 103 per mm3 20 (1/5) 31 (4/13) 15 (3/20) 0 <50 x 103 per mm3 60 (3/5) 54 (7/13) 85 (17/20) 92 (12/13) a Number of patients in the group. b Number of positive observations/number of patients examined. c An increase in the haematocrit of > 20%. report. Of the 55, 5 were classified as grade I, 14 as grade II, 23 as grade III, and 13 as grade IV; thus 36 patients (grades III and IV) had dengue shock syndrome. The pertinent signs, symptoms, and clini- cal laboratory findings in these patients are summa- rized in Table 1. The clinical symptomatology of the patients stud- ied was similar to that described in previous out- breaks (Nimmannitya et al., 1969). Of the 4 major manifestations of dengue haemorrhagic fever, fever (usually high and continuous) and hepatomegaly were found consistently in patients with all grades of the disease. Haemorrhagic manifestations, most commonly petechiae, were found in all grades except grade I. Severe bleeding-i.e., melena and or haema- temesis-was observed only in shock cases and oc- curred in 12 of the 36 severely ill patients. Haemocon- centration (a high haematocrit falling by at least 20% in the period after shock) was observed in all grades of the disease, but most frequently in grade III. The lower incidence of haemoconcentration in grade IV patients probably reflects blood loss from gastrointestinal haemorrhage. Isolation of dengue virus Dengue viruses were isolated from the serum of 9 of the 55 children at the acute stage of the disease (acute serum); 8 of the strains recovered were dengue-2 and 1 was dengue-1. The isolation of dengue viruses was correlated with the stage of the disease at which the patient was initially studied (7 of the 9 patients from whom virus was recovered were studied on or before the fourth day of the dis- ease) and with the dengue-2 HI antibody titre in the serum used for isolation. Of 16 patients with HI titres of < 1: 80 in the acute serum, 9 (56%) yielded virus; in contrast, no viruses were recovered from 39 acute sera with HI antibody titres of 1: 160 or greater. Five of the 9 patients from whom virus was recovered had both infectious virus and anti- dengue antibody in the serum at the early stage of the disease. Serological diagnosi Patients were considered to have had dengue infec- tion if 4-fold or greater rises in HI antibody titres against dengue antigens were found to occur in the serum between the acute and convalescent stages'of the disease. Patients without 4-fold rises in antibody titres were assumed to have dengue infection if the antibody titres equalled or exceeded 1: 640 during convalescence. As shown in Table 2, 45 of the 55 chil- dren had 4-fold or greater rises consistent with dengue infection; 8 additional patients had high, but fixed, antibody titres between bleedings at the acute stage and during convalescence, so they probably 120 PATHOGENESIS OF DENGUE HAEMORRHAGIC FEVER Table 2. Results of the haemagglutination-inhibition antibody test in 55 patients No. of patients with: Grade(s) of disease titre rise fixed titre no convalescent > 4-fold > 1: 640 serum tested I and 11 18 1 0 ill 19 4 0 IV 8 3 2 a Totals 45 8 2 a Both deaths. had dengue infection. Convalescent serum was not obtained from 2 patients who died with signs of dengue shock syndrome. Following the criteria of Winter et al. (1969), an attempt was made to determine what proportion of the study population had primary or secondary dengue infections. Since most of the patients were studied at least 4 days after the onset of the disease, the type of infection could not be determined accord- ing to the presence or lack of antibody in the acute serum, but was determined according to the magni- tude of HI antibody titres in the convalescent serum. Patients with HI antibody titres of 1: 640 or greater to at least 2 dengue antigens were considered to have secondary infection, whereas those with convalescent antibody titres of < 1: 640 were considered to have primary infection. By these criteria, 51 of the 53 patients whose sera were tested during convales- cence had secondary infection, as shown in Table 3. The 2 patients with primary infection were both infants (6 months of age or less) whose mothers possessed dengue HI antibody. In addition to rises in diagnostic antibody to Table 3. Numbers of primary and secondary HI antibody responses in 55 patients, according to the severity of the disease Primary Secondary Grade(s) of infection infection *- disease (convalescent (convalescent unclassifiable titre <640) titre > 640) I and 11 1 18 0 ill 0 23 0 IV 1 10 2 Totals 2 a 51 2 b a Both infants <6 months of age. b Both deaths. Convalescent serum not available. dengue antigens, 3 patients-all with grade IV dis- ease-had concurrent rises in antibody titre to chikungunya virus. The chikungunya antibody in these 3 patients was induced passively by transfusion since all had received plasma or whole blood within 24 hours of the appearance of chikungunya anti- body, and titres were at their maximum in the first serum obtained after transfusion; no further rise in titre occurred. Furthermore, chikungunya virus was not isolated from the acute serum of these 3 patients, although no chikungunya antibody was present in the early serum and the isolation method used is very sensitive for the recovery of that virus. One hypothesis that could be tested in the study was whether there was a difference in HI antibody response between children without shock (grades I and IL) and those with shock (grades lII and IV). Fig. 1 shows the log10 geometric mean dengue-2 HI antibody titres and 95% confidence limits according to the day after the onset of the disease, for both groups of patients. There were no significant differ- ences in HI antibody titre between these groups at any time after onset. It is therefore statistically per- missible to use the HI titres of patients with or with- 0 0a 0 0 C) z a 0 In z C]4 0 z 0 0 -J 5Or 4.01- = LOGIO GEOMETRIC MEAN AND 95% CONFIDENCE 1 LEVELS, PATIENTS WITH SHOCK i= LOGIO GEOMETRIC MEAN AND 95% CONFIDENCE LEVELS, PATIENTS WITHOUT SHOCK T 41+ 2.01- 4 5 6 7 8 9-10 >10 WHO 30071 DAY AFTER ONSET OF ILLNESS Fig. 1. Dengue-2 HI antibody titres following onset of disease in patients with and without shock. I 1. 121 MEMORANDA -^ ^CHILDREN WITH SERA ON DAYS 4 OR 5, 6 OR 7, AND 8 OR 9 (n=18) S.* CHILDREN WITH SERA ON DAYS 3 OR 4,5 OR 6, AND 7 OR 8 (n=10) 0 a) -0 LU u 0= zo 1- z cS 0 C14 l UJ z 0 z LI.. u y 0 I z a. 0 -J w a >) 15 zo J12 LW _6 X: a _j3 -o z WHO 30072 DAY AFTER ONSET OF ILLNESS Fig. 2. Relationship of rise in dengue-2 HI antibody titres to onset of clinical disease in two groups of patients. out shock in order to describe the antibody patterns of either group. Fig. 2 shows the geometric mean dengue-2 HI antibody titres for 2 groups of patients selected for statistical reasons according to when blood was collected for serological testing. One group of 18 patients had their serum tested on day 3 or 4, day 5 or 6, and day 7 or 8; similarly, the second group had their serum tested on day 4 or 5, day 6 or 7, and day 8 or 9. Dengue HI antibody titres in both groups rose geometrically between the 4th and 7th days after onset. It was precisely during that period that shock developed in the majority of the severely ill patients. COMPLEMENT STUDIES Normal values of immunochemically measured com- plement proteins The normal serum pool used as standard refer- ence serum was derived from apparently healthy Caucasian blood donors. The values of the reference Table 4. " Normal " serum concentration of complement proteins (,g/ml) Normal Mean of 10 Mean of 30 Caucasian different different serum pool individual individual (used as Caucasian sera Thai sera standard) ± s.d. ± s.d. Clq 130 146 ± 15 143 28 Cls 80 70+ 8 90 14 C3 1 500 1 200 ± 107 1 716 330 C4 400 611 ± 168 730 347 C5 75 75±6 71 16 C6 60 74± 8 84 18 C8 85 118± 16 107 21 C9 150 159 ± 16 183 62 C3PA 230 188 ± 35 212 63 transferrin (control) 2 500 2 863 i 43 serum are listed in Table 4 and all data given below as percentages of the normal level refer to these values. Table 4 also lists average values and standard deviations representative of 10 different Caucasian adults and of the 30 "normal" Thai children. Although there are differences between these values, it is apparent that the values of the reference serum fall within one standard deviation of the average values for the Thai sera, or very near this range. Average lowest values of complement proteins and fibrinogen for disease grades I to IV The average lowest values of Clq, Cls, C3, C3PA, C4, CS, C6, C8, C9, fibrinogen, and transferrin for all the patients (grades I-IV) are shown in Fig. 3. X 140 I 0 z 121 0 be leeis In 'J 40 20 f If I Sr. In Sr Gr. 1M Clq CIs C3 C3PA C4 C5 CC Cl CS FIRN- TRANS- OGEN FERRIN Fig. 3. Mean lowest values (± 1 standard deviation) of various complement proteins and fibrinogen from 55 individuals with serologically established dengue infection. Transferrin was measured as a noncomple- ment protein control. 122 PATHOGENESIS OF DENGUE HAEMORRHAGIC FEVER 140- Fib 12 )ihb Fib 2100/ o Fik I DO 80 F/ 20 Cr1!I be . . . . SHOCK +1 +2 +3 +4 +5 ONSET DAY 30074 WHO Fig. 4. Mean values of C3, C4, C3 proactivator, and fibrinogen from 13 patients with grade IV dengue haemorrhagic shock. The protein that exhibited the most marked diminu- tion is C3 (between 34% and 55% of normal). Also, markedly affected in grades III and IV are C3PA, C4, and C5 (about 50% of normal). For Clq, Cls, C6, and fibrinogen, the average lowest value in grades Ill and IV did not fall below 65% of normal. In contrast, the lowest values of C9 tended to be supranormal. Although transferrin was lowered, there was no difference between grades I-IV in this respect. In all other cases, the average lowest values were inversely related to the severity of the disease. Time course of C3, C3PA, and C4 in patients of grade IV Fig. 4 depicts the average values of C3, C3PA, C4, and fibrinogen in 13 grade-IV patients at the time of onset of shock and on subsequent days. All 3 factors of the complement system and fibrino- gen were low at the onset of shock and immediately thereafter rose to attain between 70% and 140% of normal values on day 5. Correlations between various complement values In Fig. 5 and 6, the lowest C3 values are com- pared with the lowest C4 or C3PA values for the 100G Iz 00Ube 3 0 8o0 60- 40- 20- 0 0 0 'A 0* *0A Ao A AA A A S P. 10 A l _'i I IFr I I[T20 40 60 80 100 LOWEST C4 ( OF NORMAL) F_al. lug" Gr. ifA 1i20 140 WHO J0071 Fig. 5. Correlation between the lowest C3 and lowest C4 values of all grade IlIl and grade IV patients (r2=0.28). The slope of the line (0.25) was determined by the least-square method. 100- * 040- 20 20 40 60 80 O60 LOWEST C3PA (X OF NORMAL) °WO 30076 Fig. 6. Correlation between the lowest C3 and lowest C3 proactivator values of all grade IlIl and grade IV patients (r2 = 0.35). The slope of the line (0.41) was determined by the least-square method. same individual. In both instances a reasonably good correlation is apparent, suggesting that both the classical (Cl, C2, C4) and alternative pathways of activation are operative in C3 depletion. The fact that the least-square line intercepts the y-axis at positive values indicates that residual C3 may be encountered when both activation mechanisms have been depleted. Fig. 7 compares the C4 and C3PA concentrations in the same serum sample for all patients of grades III and IV. Approximately 80% of the values show a 123 MEMORANDA A A A :* 0- * * * * .0 * A.-.^ * *- *-^0|- * A .* .~ " . *0 * c 0 0 A A 100- 80- 60- 40- 20- a SAME DAY Gr. Il - Gr. a 0 0 *A * A.0 * 000 A A m_TiF. 11lE Gr. 20 40 60 80 100 LOWEST TRANSFERRIN (X OF NORMAL) ° 3079 Fig. 8. Minimal correlation between lowest C3 and lowest transferrin values of all grade IlIl and grade IV patients (r2= 0.16). A 20 40 60 80 100 C3PA (% OF NORMAL) 120 140 WHO 30078 Fig. 7. Comparison of C4 and C3 proactivator con- centrations in the same serum samples of all grade IlIl and grade IV patients (r2 = 0.30). good correlation, suggesting that, in most patients, at a given time during the course of the disease, both complement activation mechanisms were opera- tive to the same extent. The residual values indicate that, in a small proportion of patients, one mechan- ism may be involved to a greater extent than the other. The correlation between platelet numbers and the lowest C3 or lowest fibrinogen concentrations and between C3 and fibrinogen values are presented in a later section (Fig. 12-14). Lack of correlation between transferrin and C3 or fibrinogen The distributions of the lowest C3 and lowest transferrin values (Fig. 8) and of the lowest fibrino- gen and lowest transferrin values (Fig. 9) are nearly circular and nearly horizontal, respectively. This dis- tribution indicates that transferrin concentrations varied independently of changes in C3 or fibrinogen. CH50 data yielded no useful information. The C2 titres often paralleled the immunochemically deter- mined C4 values and the C6 titres were generally in agreement with the C6 protein concentrations. The haemolytic titres are therefore not given in this paper. 100- t 80 z U. K 60 z 9 40 B 20-q0 0 A 0 A A 0 AA 0 A 0 0 A 0 Gr. m. Gr. 1IA 20 40 60 80 100 120 140 LOWEST FIBRINOGEN (%OF NORMAL) Fig. 9. Lack of correlation between lowest fibrinogen and lowest transferrin values of all grade IlIl and grade IV patients (r2 = 0.08). Converted C3 in EDTA plasma In the more severely ill patients, C3 conversion products could be detected by immunoelectropho- resis. Visual inspection of the immunoelectrophoresis plate showed that the degree of conversion never exceeded 25 %. Functional integrity of alternative pathway There was considerable variation in the conver- sion of C3PA to C3 activator by inulin. In general, however, the inulin test was negative or only weakly positive in early serum samples from grade IV 2 ox z LI. 0 180, 160- 140- 120- 100- 80- 60- 40- 20- 124 WHO 10080 PATHOGENESIS OF DENGUE HAEMORRHAGIC FEVER 240- 1 200- 0 ° 160 C 12 M ! to u4 40 tA 18 P1 Cl1 Cls C3 C3PA C4 C5 C6 CO Fig. 10. Mean lowest values (± 1 stan of various complement proteins and f individuals with infections other than dE measles + mumps; open circles: miscel monia, empyema, typhoid, tuberculou. haemolytic streptococcal pharyngitis). count was 27 000/mm3. Therapy included Ringer's lactate solution and whole blood given intravenously. The bar indicates the period of shock. Melena con- tinued for 2 days. The platelet count became normal (221 000/mm3) on day 7. Dengue-2 virus was isol- ,I.( ated on day 4, and the HI antibody had risen from<1 :20 to 1:160 by day 11. The child's mother j had serum antibody to all 4 dengue serotypes. The C3 and C4 levels were very low from the onset of shock (day 4) and the C3PA concentrations were less depressed but definitely affected. This situation C9FIRN- TRANS- persisted until day 10, and between days 10 and 21 OGEN FEMN a remarkable recovery of all complement proteinsidard deviation) was observed to normal or supranormal levels ngueogSquares: (C9: 200% of normal). It is noteworthy that trans- llaneous (pneu- ferrin maintained nearly the same concentration s meningitis, P- throughout the period of observation. patients and tended to become clearly positive with recovery. This finding is consistent with the view that the alternative pathway is partly or totally depleted during the acute stage of the disease in severely ill individuals. In no instance was the gamma form of C3PA or C3 activator detectable in fresh EDTA plasma. Complement profiles of other viral and bacterial infections Serial serum samples from 7 patients with mumps or measles and from 6 patients with various bac- terial infections (typhoid fever, tuberculous menin- gitis, empyema, etc.) were analysed immunochemi- cally. The results are depicted in Fig. 10. In contra- distinction to the dengue haemorrhagic fever cases, all values for this group of patients were approxi- mately normal. The average lowest values for C4 and C9 were considerably above normal. Complement changes in selected cases Patient No. 26, grade IV (Fig. 1 1A). This 6-month- old female had had a history of high fever for 3 days prior to admission to hospital, accompanied by vomiting and anorexia. Immediately before she was admitted, she became unconscious, cyanotic, and had blood in the stool. On admission (day 4) her blood pressure was not detectable, the pulse rate was >180/min, the liver was palpable 2 cm below the costal margin, and petechial haemorrhages were present; the haematocrit (Hct) was 24%, the white cell count (WBC) was 19 300/mm3, and the platelet Patient No. 32, grade IV (Fig. 1 1B). This 8-year- old Thai girl with thalassaemia-HgbH disease was admitted on the 6th day of illness with a history of fever. Her blood pressure on admission was 70/60, scattered petechiae were noted on the right arm, and the liver was palpable 2 cm below the costal margin. The Hct on admission was 27% and the platelet count 42 000. The patient was treated with a blood transfusion and fluids given intravenously, and her blood pressure became normal within 6 hours (bar). No major bleeding manifestations developed during her stay in hospital. No virus was isolated from the blood. Dengue-2 HI antibody titres were 1 :160 on day 6 and >1: 20480 on day 10. In this patient, shock set in on day 6, with a precipitous fall in the C3PA level (from 130% to 24% of normal), whereas the C4 level varied between 40% and 60% of normal. C3 was 30% of normal for 3 days, then rose and became normal on day 18. Transferrin was more or less constant during the shock period, then fell to 50% of normal while C3 was recovering. At day 18 the transferrin level was again normal, but Clq was 200% of normal. Patient No. 33, grade IV (Fig. 1 C). This 9-year- old Thai boy was admitted on the 4th day of illness with fever that continued until the 6th day, when he developed shock (bar) with undetectable blood pressure. The Hct was 39% on admission and rose to 43% on the day when shock developed, despite blood loss from haematemesis. The platelet count was 34 000/mm3 on day 4, falling to 16 000/mm3 on the day of shock. The liver was palpable 4 cm below the right costal margin on admission. Therapy includ- 9 125 A-3 ---4 --PA Tr ~ ~ ~ Ii 5 i l 21 DAY wO see iii . i DAY U. 0 z U 0 Wt 281 11 151 140 121 is, 41 2s is ft f2 Mo soes i 1 I i i DAY DAY Fig. 11. Profile of serum complement proteins and transferrin (control). 281 18 is 14 121 I, 2 13 SIF 14- 21l 2W 148 23 I U. a at m.j 0 at wo sees WmO so"e I PATHOGENESIS OF DENGUE HAEMORRHAGIC FEVER E Ilq I -I ocI a2 0 ---4 ---PA -'--3 2081 lot ISO 140 120 100 s0 51 40. 20 i i i i l's DAY VW 'Mus Fig. 11. Profile of serum complement proteins and transferrin (control) (continued). ed hydrocortisone, plasma, and whole-blood trans- fusions. The patient recovered from shock on the 9th day of the disease. Dengue-2 virus was isolated on day 4. The HI antibody titre was < 1: 20 on day 4, 1:320 on day 6, and >1:20480 on day 12. On day 5, before shock was manifest, C3 and C3PA were 70-75% of normal and C4 was above normal (150%). With the onset of shock, C3 and C4 fell, whereas C3PA was little affected. During the period of observation, transferrin and also C8 remained approximately constant, whereas Cls, C5, C6, and C9 rose sharply, remained level during the shock period, and then continued to rise to normal or supranormal values. There was a marked divergence in the concentrations of Cls and Clq, but on day 12 both became normal. Patient No. 13, grade III (Fig. 11D). This 6-year- old Thai-Chinese girl had fever for 5 days and epistaxis 2 days prior to admission. On the 6th day of illness, she was admitted with a systolic blood pressure of 50 mm Hg, a pulse of 120/min, and cold clammy skin with rare petechiae on the extremities. The liver was not palpable on admission but became palpable one day later. A tourniquet test was posi- tive. On admission, the Hct was 45%, the WBC 20 200/mm3, and a decreased platelet count was shown by smear. The patient had abdominal pain, and there was blood in her stool once during the first hospital day. The period of shock is indicated by the bar; the patient's blood pressure became normal after intravenous infusion of 1 unit of plasma and fluids. Virus was not isolated from the initial serum, and the dengue-2 HI antibody titres rose from 1: 320 to >1: 5 120. The C3 level was depressed on admission and the C3, C4, and C5 levels were very low on the following day; C3PA was not as 2100 189- 159- 149- 129 100 59- s0. 40- 20- .I ac F lq -3 __ 4 _._ PA DAY 7 5 MHO meS 127 MEMORANDA severely affected. Transferrin remained between 80% and 900% of normal and later rose. This pattern is indistinguishable from a grade IV pattern. Patient No. 41, grade II (Fig. I IE). This 10-year- old Thai boy was admitted on the 4th day of illness with a history of fever, anorexia, and abdominal pain. On admission, scattered petechiae were noted and his liver was palpable 3 cm below the right costal margin. The haematocrit on admission was 50%. Three days later, the Hct was 40%, the WBC 4 150, and the platelet count 14 000/mm3. No evi- dence of shock was manifest during the patient's stay in hospital and, apart from petechiae, there were no haemorrhagic manifestations. The child was treated with an intravenous infusion of electrolyte solution. Virus was not isolated from serum obtained on the 6th day of the disease. The serum showed an HI antibody titre of >1: 20 480 against all dengue antigens. The C3, C4, CS, C8, C3PA, and transferrin levels were initially 50-90% of normal (day 6). By day 10 all values had risen to between 840% and 153 % of normal. The studies in this case were begun quite late in the course of the infection, as evidenced by the antibody levels. Patient No. 45, grade II (Fig. 11 F). The case of this 8-year-old Thai boy followed the pattern of a shock case, although no clinical shock occurred. The child was admitted on the 4th day of illness with a history of fever, anorexia and vomiting, and abdominal pain. No petechiae were noted and the liver was palpable 2 cm below the right costal margin. The Hct was 42% on admission, rose to 500% on the 5th day of disease, and decreased to 36% on the 8th day. The platelet count on admission was 114 000/ mm3, and dropped to 12 000/mm3 on the 5th day of the disease. The child was treated with intravenous fluids. No haemorrhagic manifestations developed while the patient was in hospital. Dengue virus was not isolated from serum obtained on admission. HI antibody titres against dengue-2 rose from 1: 80 on day 4 to 1: 1 280 on day 5, and had attained > : 20480 by day 6. Between days 4 and 5, the C4 level dropped from 2000% to 65% of normal, and between days 4 and 6 C3PA fell from 110% to 60% and C3 from 85% to 28%. There was also a marked effect on CS, C6, and C9, but not on Cls and C8. However, transferrin also exhibited a marked trough at day 6, which raises the question of a general decrease in the serum protein concentration. This patient had marked haemoconcentration on day 5 and subsequent haemodilution without plasma infusions. Clq studies The results of the capillary tube precipitation test for Clq precipitins in sera are summarized here for 40 patients. For 4 grade-I patients tested, the average result for all bleedings was ± with a range from 0 to +. For 11 grade-II patients tested, the average value for all bleedings of all patients was midway between ± and + and the average maximum value for each patient was slightly better than + with extremes of 0 to + +. For the 13 grade-Ill patients studied, the average value was slightly better than + and the average maximum value for each patient was nearly ++ with extreme values of ± to +++. For the 12 grade-IV patients, the average of all bleedings was + + and the average maximum between + + and + + + with extremes of ± to +++. Whereas these preliminary results roughly reflect the severity of the disease, there was considerable variation from patient to patient within each group including the control group. Furthermore, the values for individual patients were relatively constant throughout the course of the disease and did not reflect changes in the severity of the disease from day to day. A single grade-IV patient and 2 grade-III patients were virtually negative throughout the entire course of the disease. In order to verify and extend these observations, 26 of the same sera were studied for reactivity with Clq in double diffusion reactions in gel (Agnello et al., 1970). The number of tests was limited by the lack of sera; however, 7 of the 10 sera that were positive by capillary precipitation also reacted in gel and 3 of the 5 sera that were negative by pre- cipitation did not react in gel either. The reason for the failure of agreement between the two tests is not known but is consistent with the experience of others that the reaction in gel is a more demanding and perhaps more specific test than capillary precipi- tation (H. G. Kunkel, unpublished observations). Preliminary screening of the patients' sera for reactivity with monoclonal rheumatoid factors known to react with aggregated human Ig and immune complexes (Winchester et al., 1971) showed little correlation with the severity of the disease. Random reactivity was observed in all 4 disease groups and in the control group. Furthermore, no 128 PATHOGENESIS OF DENGUE HAEMORRHAGIC FEVER close relationship between Clq precipitin reactivity and rheumatoid factor reactivity was observed. 100- 80- CLOTTING STUDIES The average fibrinogen level determined in 5 plasma samples from the Thai control group was 1.72 g/litre. The average minimum fibrinogen levels for patients in the various disease grades were: grade I (4 patients), 1.51 g/litre; grade II (14 pa- tients), 1.20 g/litre; grade III (22 patients), 0.98 g/litre; and grade IV (12 patients), 0.86 g/litre. These levels correlated roughly with the platelet levels (Fig. 12) and C3 levels (Fig. 13). The fibrinogen values in grade lV and to a lesser extent in grade III may have been increased by the blood and plasma transfusions that many of these patients received. The low fibrinogen levels seen in association with the more severe grades of dengue are all the more significant as one would expect increases in the fibrinogen levels in acute disease unless synthesis decreased or consumption increased. Split products of fibrin and fibrinogen were found in the plasma of patients with all grades of the disease. The percentages of blood samples that con- 140- 120 1001 R 80- CD 0 z 60- esU. 0 40-02 20- . A 0 . 0 a A 0 I.A A A* 0 . Gr. I[[* Gr. fA 20 40 60 80 100 LOWEST PLATELETS ( x 10-3 ) WO 30088 Fig. 12. Correlation between the lowest fibrinogen values and lowest platelet counts for grade IlIl and grade IV patients. 0 1 0 U. 0 60- 40- 20- 0 A : * :0 0 0 A~~**A *3 0I- 20 40 60 80 100 LOWEST FIBRINOGEN (%OF NORMAL) 0 Gr. mI Gr. ffA& 120 140 WPHO 30087 Fig. 13. Correlation between the lowest C3 and lowest fibrinogen values for grade IlIl and grade IV patients. tained split products were: grade I, 37%; grade 11, 41 %; grade 1II, 51 %; and grade IV, 53 %. The aver- age amounts of split products (mg/100 ml) in all blood samples were: grade I, 1.7; grade II, 3.2; grade III, 2.9; and grade IV, 5.1. The frequent occur- rence of split products in the plasma of patients with all grades of the disease suggests that the con- sumption of fibrinogen is an intrinsic part of the disease process. Platelet counts were performed in the laboratories of the hospitals in which the patients were treated. The average platelet counts in all bleedings accord- ing to the disease grade were: grade I, 108 000; grade II, 82 000; grade III, 45 000; and grade IV, 60 000. The value in grade IV may reflect the repeated whole-blood transfusions given to most of these patients. The average lowest counts recorded for patients in the various grades were: grade I, 59 000; grade 1I, 49 000; grade III, 24 000; and grade IV, 20 000 (omitting one unusual patient in the latter group), and correlated roughly with the fibrinogen levels (Fig. 12) and C3 levels (Fig. 14). Again, the reductions in the platelet levels with the severity of the disease suggest that platelets were involved in the disease process. IMMUNOFLUORESCENCE STUDIES Peripheral blood leucocytes from 35 patients se- lected from all grades of the disease were stained in living suspensions (Rabellino et al., 1971) in the hope of finding abnormal proteins on their surfaces. The peripheral leucocytes did not stain for C3, 129 MEMORANDA 100 J IL 0 0 at uU) 0 _1 80- 60- 40- 20- 0 *-* * A 20 40 . 6o 5o LOWEST PLTELTS (x 10-3 ) Fig. 14. Correlation between the lowest C: lowest platelet counts for grade Ill an patients. fibrinogen, albumin, or dengue antigen globulin (Ig) staining of lymphocytes i normal granular and cap patterns in t patients and the control group. Howc staining of granulocytes from patients dengue revealed significantly more surf was present on control granulocytes. was some staining of even control granu since the granulocytes in persons with are no doubt quite different from the E in healthy individuals, it is difficult to e observation. However, the recent demo. Ig receptors on granulocytes (Henson, 1 sistent with this observation and suggesi receptors play a role in the disease. DISCUSSION The 55 patients included in this study manifestations of the disease caused by c infection, 36 of them showing the dengu( drome. The clinical manifestations in all were characteristic of dengue haemorn (Nimmannitya et al., 1969) and the p; shock exhibited the pathophysiological ch; ing from increased vascular permeabilit, Halstead, 1966). The 53 patients for who serological data were obtained showed evidence of current or recent group I infection, and only dengue viruses wo from these patients. Since infection by another group B arbovirus is highly unlikely in the Greater Bangkok area, these findings confirmed the dengue virus etiology of the disease. Viraemia was not commonly found in the patients studied, and indeed was detected only in the pres- ence of low dengue antibody titres. Failure to detect dengue virus in patients with initial HI antibody levels of 1: 160 or greater is probably a reflection of in vivo neutralization of the dengue virus by the specific antibody. It is possible that viral antigen may have Gr. II[* been present in the serum of patients with high Gr. f A antibody titres yet was not revealed by the detection 10 . method that relies on demonstration of the presence 10lO0 of unneutralized infectious virions. Virus was found in the serum of 5 patients who also had small 3 values and amounts of reactive antibody in the same serum id grade IV specimen; thus the basic conditions for immune complex formation were present in these patients. Of the 55 patients studied, 53 had secondary dengue infections and a rapid geometric increase in s. Immuno- dengue-2 HI antibody was observed in those from revealed the whom adequate serial blood samples could be *oth dengue obtained. The 2 patients with primary dengue infec- ,ver, the Ig tion were the infants of mothers with dengue anti- with severe body and presumably had acquired maternal dengue ace Ig than IgG placentally at the time of dengue virus infection. Since there Indeed, with the exception of one of these 2 infants, ocytes, and all patients developing dengue shock syndrome had the disease dengue antibody in the serum obtained during shock, ;ranulocytes and most showed a rapid rise in antibody in the valuate this subsequent few days. The data obtained are thus nstration of consistent with the hypothesis that dengue shock 971) is con- syndrome occurs when dengue antigens and anti- s that these body are present simultaneously-a situation in which the in vivo formation of complement-fixing immune complexes is likely. Direct proof of the presence of circulating immune complexes was not obtained during the study. The nature of the Clq reactive material found is not had severe known, nor is its relationship to the pathogenesis lengue virus of the dengue haemorrhagic shock syndrome clear. shock syn- The fact that rheumatoid factor did not react with 55 patients most sera that precipitated with Clq would suggest rhagic fever that the Clq was not detecting conventional antigen- atients with antibody complexes. However, the sensitivity of both anges result- Clq and rheumatoid factor precipitin tests is limited. y (Cohen & Whether the Clq reactivity was caused by DNA, fm adequate endotoxin, or perhaps the products of virus or virus serological tissue interaction is the subject of continuing study. 3 arbovirus Low platelet counts, reduced fibrinogen levels, the ere isolated presence of circulating split products, and the prelimi- w X w . w - -130 . PATHOGENESIS OF DENGUE HAEMORRHAGIC FEVER nary evidence of reduced Hagemann factor (C. G. Cochrane, unpublished observations) together pro- vide strong evidence of disseminated intravascular coagulation in the dengue haemorrhagic shock syn- drome. This is consistent with the earlier observation of Fresh et al. (1969). The fact that the clotting mechanism is apparently involved early even in the milder forms of the disease and that it is increasingly involved in the more severe forms suggests that this process is an intrinsic part of the disease and not merely a secondary phenomenon of severe shock. In most patients from whom multiple serial blood samples could be obtained, the changes of comple- ment were synchronous with those of fibrinogen and fibrin split product concentrations and platelet num- bers. Furthermore, these changes were well corre- lated in time with the onset of the haemorrhagic shock symptoms and with their severity. The increased immunoglobulin found on the sur- face of the granulocytes is not readily explainable. It is possible that immunoglobulin was present in abnormal forms-i.e., complexes-in dengue patients and, therefore, that it was picked up by the granulo- cytes; on the other hand, the granulocytes in such persons may pick up more normal Ig than the granulocytes of healthy persons do. In the absence of staining of granulocytes for C3 or dengue anti- gens, the uptake of circulating antigen-antibody complexes could not be established. The most significant results of this study relate to the complement system. Convincing evidence of in vivo complement consumption was obtained and the marked depression of serum levels of comple- ment components was found to be correlated with the severity of the disease. This is in contrast to other viral diseases, such as measles and mumps, and to a variety of bacterial infections, in which normal or supranormal complement values have been found. Normal C3 levels have been observed in patients with chikungunya, rubella, and primary dengue infections (Russell et al., 1969). Complement depression in dengue haemorrhagic fever patients involves primarily C3, C3 proactivator, C4, and C5. Considering the normal half-life of disappearance from the plasma of complement proteins (50-60 hours), the rate at which the concentration of these proteins decreased in patients with shock argues strongly in favour of the hypothesis of complement activation and consumption. The good correlation observed between the lowest C3 values and lowest C4 or C3PA values supports the concept that C3 was consumed through activation of the classical and alternative pathways. Furthermore, C3PA and C4 were usually similarly depressed in the same serum sample. But there were also exceptions with supranormal C4 values and low C3PA concentra- tions, and vice versa, suggesting that in some cases one of these mechanisms of activation was involved to a greater extent than the other. The relatively stable transferrin levels show that the depletion of complement proteins was not pri- marily due to extravasation. No correlation was found between the lowest C3 and lowest transferrin values. There remains a possibility that extravasation and decreased synthesis of complement components may be partly responsible for the low levels. Although this seems unlikely in view of the precipitous de- creases observed, studies of complement metabolism will be required to evaluate this possibility. What might be the pathological consequences of a massive activation of complement? Activation of C3 and C5 is accompanied by dissociation of low molecular weight peptides called C3a and C5a, which have the capacity to release histamine (MUller- Eberhard & Vallota, 1971) and are very potent per- meability increasing factors. The circulation of an individual weighing 20 kg contains 1.2 g of C3, which can potentially release 44 mg of C3a or 6 million minimal weal and flare doses (tested in the human skin). (It is noteworthy that C3 levels in the shock cases drop to as low as 20% of normal.) From the amount of C5 present in such an indi- vidual's circulation, potentially 4.4 mg of C5a can be released, corresponding to 28 million minimal weal and flare doses. Since the C3 concentration was reduced by at least 33% in the shock patients and a 33% reduction in the C5 levels also occurred in 89% of such patients, it may be that large amounts of anaphylatoxins are liberated during the shock phase. Although plasma contains a powerful inacti- vator of the two peptides (Bokisch & Muller- Eberhard, 1971), it is quite possible that these pep- tides contribute to the development of the shock before they undergo inactivation. Complement has been shown to be capable of initiating blood coagulation through the platelets (Zimmerman & Miiller-Eberhard, 1971). Platelets acquire coagulant activity particularly through acti- vation of the bypass mechanism. Thus, at least in part, the disseminated intravascular coagulation observed in dengue haemorrhagic fever may be a consequence of intravascular complement acti- vation. A plausible working hypothesis, therefore, is as 131 MEMORANDA follows. As a result of a dengue virus infection in a patient with pre-existing antibody, an immuno- pathological process possibly involving immune complexes or another Clq reactive substance leads to massive complement activation. Activation is accompanied by the liberation of vasoactive pep- tides (anaphylatoxins) and by initiation of intra- vascular blood coagulation, followed by partial, often severe, depletion of plasma complement pro- teins owing to the rapid removal of complement reaction products from the circulation. The simul- taneous activation of proteolytic enzymes of the complement, coagulation, and possibly the kinin, systems may be expected to consume plasma enzyme inhibitors. Partial depletion of these inhibitors may result in an imbalance between activated enzymes and inhibitors and thus produce increased vascular permeability and shock. The central role played by the complement system in producing vascular permeability changes appears to have been established. The activation of the alter- native pathway as well as the classical pathway was an unexpected finding and is the first association of this mechanism with an immunopathological dis- ease. The extremely complex interactions between the complement system, the kinins, the clotting system, and platelets remain incompletely under- stood and will require additional investigation. iUMI MECANISMES PATHOGENIQUES DE LA DENGUE/FItVRE HEMORRAGIQUE: RAPPORT D'UNE ETUDE COLLECTIVE INTERNATIONALE Cinquante-cinq patients atteints du syndrome dengue/ fievre hemorragique, dont 36 pr6sentant la forme la plus grave de la maladie avec etat de choc, ont fait l'objet d'investigations intensives. Le but principal de cette etude etait d'6tablir les relations entre l'6volution clinique, les caracteristiques hematologiques, virologiques et sero- logiques d'une part, et les modifications des constituants du complement et les troubles immunopathologiques d'autre part. On a isole des virus de la dengue des types 1 ou 2 des serums preleves 'a la phase aigue chez 9 malades. Chez 51 patients, les anticorps anti-dengue etaient presents a des titres indiquant une infection secondaire par le virus de la dengue (titres IH > 1: 640); chez deux autres, les titres signaient une infection primaire (titres IH < 1: 640). Les titres se sont eleves suivant une progression geome- trique pendant la phase aigue de la maladie. On notait une diminution marquee des concentrations seriques des constituants du complement, en particulier de C3. La depletion de C3 semblait resulter d'un processus d'acti- vation empruntant la voie classique ou la voie de substi- tution; les deux mecanismes intervenaient generalement dans une mesure semblable, bien que dans certains cas l'un ou l'autre ait ete davantage sollicite. Les taux seriques des constituants du complement etaient d'autant plus bas que l'affection 6tait grave. Quant aux taux des trans- ferrines, leur relative stabilite indiquait que la depletion des proteines du complement n'etait pas due en ordre principal a un phenomene d'extravasation. Les taux de fibrinogene etaient inferieurs a la normale et le plasma contenait des fractions de la fibrine et du fibrinogene. On peut des lors consid6rer comme plausible l'hypo- these ci-apres. A la suite d'une infection par un virus de la dengue chez un sujet deja porteur d'anticorps, un processus immunologique faisant intervenir des immun- complexes ou, peut-etre, une autre substance Clq-reactive aboutit a une activation massive du complement. Cette activation est accompagnee de la liberation de peptides (anaphylatoxines) agissant sur les vaisseaux et d'un debut de coagulation sanguine intravasculaire, puis d'une deple- tion partielle, souvent marquee, des proteines du comple- ment due a l'elimination rapide de la circulation des pro- duits de reaction du complement. I1 est probable que l'activation concomitante des enzymes proteolytiques des syst6mes du complement, de la coagulation et peut-etre des kinines entraine un epuisement des inhibiteurs de ces enzymes. La carence partielle des inhibiteurs provoquerait un des6quilibre entre les enzymes activees et les inhibiteurs, avec comme consequence une permeabilit6 vasculaire accrue et un etat de choc. Le r6le central joue par le systeme du complement dans le declenchement des troubles de la permeabilite vascu- laire parait bien etabli, mais les interactions complexes entre ce systeme, les kinines, les facteurs de coagulation et les plaquettes ne sont pas encore completement eluci- dees. 132 PATHOGENESIS OF DENGUE HAEMORRHAGIC FEVER 133 REFERENCES Agnello, V. et al. (1970) Immunology, 19, 909-919 Arroyave, C. M. & Muller-Eberhard, H. J. (1971) Immunochemistry, 8, 995-1006 Bokisch, V. A. & Muller-Eberhard, H. J. (1971) J. clin. Invest., 49, 2427-2436 Calcott, M. A. & Muller-Eberhard, H. J. (1972) Bio- chemistry, 11, 34-43 Clarke, D. H. & Casals, J. (1958) Amer. J. trop. Med. Hyg., 7, 561-573 Cohen, S. N. & Halstead, S. B. (1966) J. Pediat., 68, 448-456 Cooper, N. R. et al. (1970) Immunochemistry, 7, 341-356 Edgington, T. S. (1972) Radial diffusion method for fibri- nogen and split products. In: W. J. Williams, ed., Hematology, New York, McGraw Hill, p. 1410 Fresh, J. W. Reyes et al. (1969) J. Lab. clin. Med., 73, 451-458 Gotze, 0. & Muller-Eberhard, H. J. (1971) J. exp. Med., 134 (Suppl.), 90s-108s Hadding, U. & Muller-Eberhard, H. J. (1969) Immunol- ogy, 16, 719-735 Haines, A. L. & Lepow, I. H. (1964) J. Immunol., 92, 456-467 Halstead, S. B. et al. (1967) Jap. J. med. Sci. Biol., 20 (Suppl.), 96-102 Henson, P. M. (1971) J. exp. Med., 134 (Suppl.), 114s-135s Mancini, G. et al. (1965) Immunochemistry, 2, 235-254 Manni, J. A. & Muller-Eberhard, H. J. (1969) J. exp. Med., 130, 1145-1160 Muller-Eberhard, H. J. & Biro, C. E. (1963) J. exp. Med., 118, 447-466 Muller-Eberhard, H. J. & Vallota, E. H. (1971) Forma- tion and inactivation of anaphylatoxins. In: Proceedings of the Second International Symposium on the Bio- chemistry of the Acute Allergic Reaction, Augusta, Mich., 1971, Oxford, Blackwell, p. 217 Nilsson, U. & Muller-Eberhard, H. J. (1965) J. exp. Med., 122, 277-298 Nimmannitya, S. et al. (1969) Amer. J. trop. Med. Hyg., 18, 954-971 Rabellino, E. et al. (1971) J. exp. Med., 133, 156-167 Rother, K. et al. (1966) J. exp. Med., 124, 778-785 Russell, P. K. (1970) Pathogenesis of the dengue shock syndrome: evidence for an immunologic mechanism. In: Proceedings of the Sixth International Symposium on Immunopathology, Grindelwald, Switzerland, 1970, Basle, Schwabe, pp. 426-435 Russell, P. K. (1971) Immunopathologic mechanisms in the dengue shock syndrome. In: Progress in immunology, New York, Academic Press, pp. 831-838 Russell, P. K. & Nisalak, A. (1967) J. Immunol., 99, 291-296 Russell, P. K. et al. (1969) J. Immunol., 102, 412-420 Wilson, C. B. & Dixon, F. J. (1970) J. Immunol., 105, 279-290 Winchester, R. J. et al. (1971) J. exp. Med., 134 (Suppl.), 286s-295s Winter, P. E. et al. (1968) Amer. J. trop. Med. Hyg., 17, 590-599 Winter, P. E. et al. (1969) Amer. J. trop. Med. Hyg., 18, 573-579 Yuill, T. M. et al. (1968) Amer. J. trop. Med. Hyg., 17, 441-448 Zimmerman, T. S. & Muller-Eberhard, H. J. (1971) J. exp. Med., 134, 1601-1607 Annex INSTITUTIONS PARTICIPATING IN THE COLLABORATIVE STUDY Chulalongkorn Hospital, Bangkok, Thailand (Dr Chulee Mitrakul & Dr Prasit Futrakul) Faculty of Medicine, Ramathibodi Hospital, Bangkok, Thailand (Dr Aree Vulayasevee, Dr Boonchob Pong- panich, Dr Natth Bhamarapravati, Dr Pattraporn Bhanchet, & Dr Phaiboolya Phanichyakarn) Faculty of Public Health, Mahidol University, Bangkok, Thailand (Mrs Anong Pariyananda & Dr Suchinda Udomsakdi) Faculty of Science, Mahidol University, Bangkok, Thailand (Dr Stitaya Sirisinha) Scripps Clinic & Research Foundation, La Jolla, Calif., USA (Miss Mary Ann Calcott, Miss Patricia J. McConahey, Dr Frank J. Dixon, & Dr Hans J. Muller- Eberhard) Siriraj Hospital, Bangkok, Thailand (Dr Prasong Tu- chinda, Dr Sombodhi Bukkavesa, & Dr Vinai Suwatte) US Army Medical Component, SEATO, Bangkok, Thailand (Dr Franklin H. Top, Jr) Walter Reed Army Institute of Research, Washington, D.C., USA (Dr Philip K. Russell) Women's & Children's Hospital, Bangkok, Thailand (Dr Pethai Mansuwan & Dr Suchitra Nimmanitya) 10