Reader Comments

Post a new comment on this article

The existing pre-1981 serological evidence singles out the Democratic Republic of Congo

Posted by jdsousa on 05 May 2010 at 14:01 GMT

The review paper written by Rémy (1993) lumps Kinshasa and Brazzaville together in its abstract, because he is referring mostly to the epidemiological situation of the late 1980s. Almost all serological studies he reviews are from the period 1985–91. At this time, both the Democratic Republic of Congo (DRC) and Congo-Brazzaville had similar HIV-1 prevalences (around 5%).

However, if we concentrate on the serological studies from before 1981, part of which are listed also in Rémy (1993) we get a different reality. Here is the list of all serological studies that we are aware of with sera collected from Central Africa before 1981:

Country, place, time: number HIV-1+/number bled (% prev)

DRC, Kinshasa and rural Congo, 1959: 1/672 (0.15%) (Note 1)
DRC, Kinshasa, 1967–70: 2/805 (0.25%)
DRC, Kinshasa, 1980: 15/498 (3.01%)
DRC, Mbandaka, 1969: 0/250 (0%) (Note 2)
DRC, Yambuku, 1976: 5/659 (0.76%)
Burundi, Bujumbura and other towns, 1980–81: 29/658 (4.41%) (Note 3)
Chad, Unknown location, 1974: 0/100 (0%)
Several countries, 1964–75: 0/677 (0%) (Note 4)
Sudan, Nzara and Maridi (near DRC), 1976: 0/461 (0%) (Note 5)
Gabon, Unknown location, 1967: 0/80 (0%)
Gabon, Haut-Ogooué, 1980–81: 0/437 (0%)
Gabon, Unknown location, 1975: 0/69 (0%)
Congo Brazzaville, north, 1975–78: 0/340 (0%) (Note 6)

The references are listed in the Notes at the end of this text, for the studies not listed in Rémy (1993).

We can see from the list above that, of five serological studies with sera collected from the DRC, four show evidence of HIV-1. Burundi had a high HIV-1 prevalence in 1980–81, but there is evidence that this focus was recent and secondary to the DRC focus because: 1) Unlike in Yambuku in 1986, no children were found infected in Burundi in 1980, and many were tested (Nzilambi et al 1988; Morvan et al 1989); 2) Unlike the very high viral diversity found in the DRC (Vidal et al 2000, 2005), the HIV-1 epidemic in Burundi is overwhelmingly dominated by subtype C (Vidal et al 2007).

The remaining serological studies do not show HIV-1 at all. This is particularly interesting concerning the four studies with samples collected from Gabon and Congo-Brazzaville, two countries that border south Cameroon, where the SIVcpz closest to HIV-1 group M exists in wild chimpanzees (Keele et al 2006). In addition, the Rémy (1993) review lists additional serological studies from Gabon, Equatorial Guinea, and south Cameroon, from the period 1982–86, some of them with bigh sample sizes, and the HIV-1 prevalences are zero in several of them, and lower than 0.7% in most.

Therefore, the serological studies, the table of early AIDS cases (Hooper 2000) and the distribution of subtypes, all suggest that the epicenter of HIV-1 group M was developing almost exclusively within the DRC, in the period intermediate between the decades before the pandemic “big bang” (that is, in the three or four decades preceding 1975 or so). In the countries to the west of DRC, no early HIV. The DRC seems to have exported HIV-1 more easily to its east and south then to its west.

Note 1. The initial sampling involved over 1,800 people from Kinshasa, Kisangani, Équateur province, and other regions of DRC; of these, 672 were later tested for HIV, and one single man from Kinshasa was found seropositive (Nahmias et al 1986).

Note 2. See Dube et al (1994).

Note 3. The prevalence was 8% in Bujumbura (Morvan et al 1989).

Note 4. Although most of these samples were collected in Uganda and Tunisia, a few were from DRC and Cameroon (Levy et al 1986).

Note 5. The sampled people were well characterized in Bowen et al (1978). Tests for HIV were performed later, and despite conflicting statements about the results, Françoise Brun-Vézinet claims that no HIV was found (Hooper 2000).

Note 6. These 340 people were Pygmies; a survey of 277 non-Pygmies in the nearby Sangha area in 1982 did not found HIV either (Rémy 1993).

References (the remaining references cited here are listed in the paper itself)

Bowen et al (1978) In Pattyn (editor) Ebola virus hemorrhagic fever. Amsterdam: Elsevier/North Holland Biomedical Press, p.143–51.
Dube et al (1994) Virology 202: 379–89.
Levy et al (1986) Proc Natl Acad Sci USA 83:7935–37.
Morvan et al (1989) Bull Soc Path Exot Fil 82: 130–40.
Vidal et al (2007) AIDS Res Hum Retroviruses 23: 175–80.

No competing interests declared.

RE: The existing pre-1981 serological evidence singles out the Democratic Republic of Congo

jmoore replied to jdsousa on 05 May 2010 at 16:29 GMT

I wrote: "HIV has been found in early samples from Kinshasa, but unless there's been comparable search effort in samples of comparable number and quality elsewhere, it is premature to e.g. distinguish Kinshasa from Brazzaville with respect to the disease's origin."

In a response, you wrote: " If we focus on older serological collections, another picture emerges, singling out the Democratic Republic of Congo as a far more important (and possibly exclusive) epicenter in the decades preceding the "pandemic big bang" of the mid 1970s."

And then present a very helpful table of early search effort in this comment.

Question: is the early rate of HIV infection significantly greater in Kinshasa/DRC than in Brazzaville/RPC or other countries near the suspected origin in SE Cameroon?

Data: the rate in the three 1980 samples is 1 in 36, compared with 1 in 430 for earlier samples; since the question here is about the origin of the disease, I think the 1980 surveys should be excluded (though they don't alter my point).

Start with pre-1970 data. For DRC, the rate is 1 in 576 (1 in 492 if restricted to Kinshasa). For RP Congo and Gabon, the rate is 0 in 80. Those are not significantly different.

Next, pre-1980 data. For DRC, the rate is now 1 in 298. For Congo and Gabon, it is 0 in 489. That difference might look suspicious, but there's a problem: the entire RP Congo sample comes from Pygmies; without them the rate is 0 in 149. I don't know enough about the status of Pygmies in the colonial period to argue one way or the other about whether they should be lumped in with other groups in this kind of analysis.

The 1982 survey of 277 non-Pygmies in the Sangha area which found no HIV is probably the strongest indication of a difference in the rates. EXCEPT: in Remy (1993) Table 1, it looks like the two RP Congo samples (this, and the Pygmy sample) were both taken from the general population; all of the Kinshasa samples were from hospitals or pregnant women or blood donors (pregnant women are obviously sexually active, and so are a biased sample with respect to STDs).

So the evidence supporting Kinshasa but not Brazzaville as the epicenter for the origin of HIV-1 consists of non-significant differences in rates of infection, derived from samples biased in favor of finding infections from Kinshasa, and with no samples at all from Brazzaville itself.

No competing interests declared.

RE: RE: The existing pre-1981 serological evidence singles out the Democratic Republic of Congo

jdsousa replied to jmoore on 24 Aug 2010 at 02:42 GMT

Thank you again, Jim, for your cogent comments. You are right to point out that the small number of seropositives found in the samples collected before 1980 does not permit to conclude for any differences in HIV-1 prevalence between countries or between cities, with statistical significance. Moreover, the set of serological studies listed above is, of course, biased in favour of Kinshasa.

However, we believe that the DRC was the main, and perhaps exclusive, epicenter of the epidemic in the pre-pandemic period (the few decades prior to the pandemic) by a confluence of reasons, which we already expounded in this and other commentaries, and that include, in addition to the serological evidence, the list of early AIDS cases (Hooper 2000), and the very high viral diversity found in the DRC and in Kinshasa in particular.

One of us (Philippe Lemey) is working on a phylogeographic study of HIV-1 group M in the DRC that attempts to reconstruct the historical patterns of geographical dispersal. This study includes samples from Brazzaville, and preliminary findings point at Kinshasa as the epicenter.

Here, I will add an additional argument in favour of Kinshasa versus Brazzaville (and in favour of the DRC versus Congo Brazzaville). Let us focus on the times of “take-off”, or strong epidemic growth (that is, a spread starting with small or negligible prevalence) of HIV-1 in several of the main Central African cities.

In Kinshasa, clearly the prevalence surpassed 1% during the 1970s. The prevalence was 3% in 1980, and arrived to 5% soon after, stabilizing more or less at this value.

An examination of the 170+ serological studies listed in Rémy (1993) shows that HIV-1 took off much later in the main Cameroonian cities. Focusing on the studies targeting the general population (pregnant women, blood donors, adults in general), and excluding studies targeting commercial sex workers, we can find 21 studies with sera collected from many thousands of people from Yaoundé. None is from before 1987, most are from the period 1987¬–89. The prevalences are almost all in the range 0.3–1.3%, with only two being around 2%. For Douala, we can find 9 studies, also involving many thousands of bled people, also all from 1987 or after. The prevalences are all up to 1.1% (except in a few studies which only used Elisa tests without confirmatory tests). A few years later, the prevalences stabilized at 5–6% in these two cities. Clearly, the main take-off of HIV-1 in these cities occurred in the late 1980s. For the other Cameroonian cities (Bamenda, Ngaoundéré, etc) the pattern and timings are similar.

For Libreville, Gabon, Rémy lists 6 relevant studies, all from 1986 or after. The one from 1986 shows a 1.8% prevalence; the others, show prevalences in the range 0.4–2.7%. Later, the prevalence would go to 4–5%.

Thus, HIV-1 took off in the cities of two of the main countries to the West of the DRC, Cameroon and Gabon, many years later than in Kinshasa. This conclusion is statistically very significant, because the samples involve many thousands of people, and the variation in prevalence between studies made in the same place in a short period is small. Equatorial Guinea is similar to Cameroon and Gabon.

For Bangui, Central African Republic (CAR), Rémy lists 15 relevant studies. In one, blood was collected in 1985, and the prevalence was 2.1%. In two others, blood was collected in 1986, and the prevalences were 4.0 and 4.7% respectively; being the sample size very small in the former study, the weighted prevalence of these two studies is 4.1%. The remaining studies collected blood in 1987 or after, and the prevalences are 5.2–15%, with many in the range 7–9%. This suggests that HIV arrived earlier to Bangui than to Cameroon and Gabon, and it crossed the 1% prevalence not many time before 1985. The examination of other studies from other CAR cities reinforces this, because the prevalences in the years 1985–86 are all below 2.5%, exploding in the subsequent years.

For Rwanda and Burundi, Rémy does not list studies. However, HIV took off in these countries much earlier than in the countries to the West of DRC. Many published papers and books review the evidence of the high epidemic spread there in the early 1980s. Already in 1983, an international team started to study AIDS in Kigali, Rwanda, and calculated a very high incidence of the disease. And Morvan et al (1989) found a 4.4% HIV-1 prevalence in Bujumbura, Burundi, in 1980–81 (that is, higher than the one found in Kinshasa at that time).

Now, let us go to the last bastion: Brazzaville. Sadly, it is difficult to determine the time when HIV arrived to this city, because all the 17 studies about the city listed by Rémy are from 1986 onwards, and show similar prevalences, without a clear trend. Two studies from 1986 show 4 and 4.6%. The remaining studies show prevalences in the range 3.1–6.4%. These prevalences are close to the maximum attained. From this information, it is impossible to estimate when did HIV cross the 1% mark in Brazzaville. By 1986, the prevalence was nearly the same as in Kinshasa, and flat. Neither can we draw any more conclusions from the examination of the serological studies from other cities and towns of Congo Brazzaville.

Despite this, considering that: 1) HIV-1 group M clearly took off much later in Cameroon, Gabon, and Equatorial Guinea (in the late 1980s); 2) It did so in the mid-1980s in Bangui and the rest of CAR; 3) There were always many trade and other connections between Bangui and Brazzaville; 4) In 1980, the prevalence was 3% in Kinshasa and 4.4% in Bujumbura, a city far to the East, and connected with Central Africa through the DRC only; then, it is much more natural to assume that the main epicenter was Kinshasa and the DRC in the pre-pandemic period. It seems more probable that the epidemic started in Brazzaville relatively late, perhaps in the late 1970s or early 1980s, some years before 1984–85, when it started in Bangui (a city closely connected to Brazzaville). If we postulate that the epidemics of Kinshasa and Brazzaville were proggressing simultaneously (e.g., both arriving to a 3% prevalence in 1980, and having started many decades before in both cities) it would be difficult to explain the late arrival to Bangui, and even later arrival to Cameroon and Gabon.

No competing interests declared.