New ZealandNorwayPakistanPeruPolandPeurto RicoRomaniaRussiaSingaporeSouth AfricaSudanSwedenThailandTurkeyUnited KingdomUSAUzbekistan

Advances in Inflammation Research, Vol. 9: The Spondyloarthropathies, edited by M. Ziff and S. B. Cohen. Raven Press, New York C 1985.

Etiopathogenesis of Ankylosing Spondylitis and the Cross-Tolerance Hypothesis

A. Ebringer, M. Baines, M. Childerstone, M. Ghuloom, and T. Ptaszynska

Immunology Unit, Department of Biochemistry, Queen Elizabeth College, University of London, and Department of Rheumatology, The Middlesex Hospital, London Wl, England

During the past 10 years, our group has been investigating the role of genetic and environmental factors in immune responses and in HLA-linked diseases, especially Spondyloarthropathies. A recurrent theme has been that the genetic marker is directly involved, not only in immune responses in inbred mice, but also in the pathogenesis of HLA-linked diseases in man (27).

The simultaneous demonstration in 1973 by two groups, one from London and the other from Los Angeles, that HLA-B27 is present in 96% of patients with ankylosing spondylitis (AS), whereas its frequency in the general population is less than 10%, has become a central and dominant problem in the immunogenetics of rheumatic diseases (11, 84).
HLA-B27 is also increased in Reiter's disease (12), acute anterior uveitis (AAU) (13), and especially in the reactive arthritis following gut infections by gram-negative bacteria (1).


It has been suggested by us that the various conditions in which an increased frequency of HLA-B27 is observed, such as AS, Reiter's disease, acute anterior uveitis, radiological sacroiliitis, symptomatic backache, and gram-negative reactive arthritis, be considered as different manifestations of the same syndrome, which could be called "B27 disease" when it occurs in individuals carrying the marker (30).

klebsiella absorbed

Fig 10. Comparison of serum IgA absorption (mean +- SE) with Klebsiella microoganisms in ankylosing spondylitis patients and controls, having high (above 8 g/dl) and low (below 8 g/dl) total serum proteins.

The implications of this concept are that there are two types of AS. First, there is AS that occurs in individuals carrying HLA-B27 and forms part of B27 disease, and second, there is AS that occurs in non-B27 individuals, which would appear to be a different or milder form of AS. In our Ankylosing Spondylitis Clinic, at the Middlesex Hospital in London, the dozen or so AS patients not carrying HLA-B27, followed over a period of 8 years, were found to be rarely in an active phase of the disease when this was assessed either clinically or by elevations in acute phase reactants such as C-reactive protein (CRP) or erythrocyte sedimentation rate.
The incidence of rheumatic fever has declined in the Western world, possibly through the early and adequate treatment of streptococcal infections with penicillin, and perhaps a similar approach could be explored in AS.


We have carried out two long-term antibiotic trials, each lasting 12 months, using co-trimoxazole and phthalyl-sulphathiazole in an attempt to reduce the level of acute phase reactants by modulating or changing the gram-negative bowel flora.

TABLE 2. Pathogenesis of AS, according to the cross-tolerance hypothesis


· By Klebsiella and other gram-negative microorganisms in the colon.
*"Produced against the bacteria, has also some anti-B27 activity and will bind to such antigens, especially after the infection starts to wane.
**Occurs as a result of complement activation, which may lead to tissue damage at sites, such as spine and uvea, that are different from the original sites of infection.
"Will be produced because recurrent waves of infection will lead to recurrent spinal inflammation, terminating in the chronic sequelae of established AS.

Both trials were unsuccessful and to some extent not unexpected, since resistance to antibiotics develops rapidly through plasmid R-factors in gram-negative bacteria.

Another approach was deemed desirable and review of the literature indicated that mono- and disaccharides were present in significant quantities (5-10 g/day) in ileostomy fluids (67). Furthermore, nearly all normal subjects fail to absorb an appreciable portion of starch in all-purpose wheat flour, commonly found in bread or pasta, as assessed by breath hydrogen concentration, because of an interaction between the carbohydrate and protein moieties of wheat (5).

In vitro studies were carried out to determine the relative efficiency of equal amounts (5 g/liter) of simple sugars and amino acids to act as substrates for the growth of Klebsiella pneumoniae in 24-hr cultures (20 ml Pirt's PI medium) (73).

The mean number of microorganisms per gram of substrate following culture with three different sugars was 8.7 ± 1.0 (mean ±SE)(x 109) compared to 0.8 ± 0.2 (xl09) (p<0.001) obtained following incubation with 11 different amino acids (Fig. 11).
The in vitro study led to the expected conclusion that, weight-for-weight, simple sugars are better substrates for bacterial growth than components of proteins. Thus a high protein-low carbohydrate diet might reduce the gram-negative bowel flora and this, in turn, could affect the inflammatory phase of the disease which could be followed by measuring the level of acute phase reactants.

A simple high protein-low carbohydrate diet was prepared, the emphasis being that patients should reduce their consumption of carbohydrates and increase their intake of protein and vegetables. The high protein-low carbohydrate diet was tested for 8 weeks in 8 non-B27 healthy volunteers and this led to a mean serum IgA fall of 35.4 ± 12.9 mg/dl (p<0.05). The CRP level in all 8 volunteers at the start of the trial was zero. Thus serum IgA tended to decrease in normal individuals following such a diet, but the numbers of subjects are too small to warrant drawing any conclusions.

Mean number of Klebsiella pneumoniae microorganisms following 24-hour culture using different carbohydrates and amino acids as substrates

FIGURE 11: Mean number of Klebsiella pneumoniae microorganisms following 24-hour culture using different carbohydrates and amino acids as substrates.

An open trial of the same diet was carried out in 36 active (ESR>15 mm/hr) AS patients over a period of 9 months (Fig. 12). The mean ESR at the onset of the trial was 38.8 ± 3.1 mm/hr and at the end it was 24.3 ± 2.4 mm/hr (t = 6.76, p<0.001), whereas no such difference was observed in the same group of patients when examined retrospectively for the same period of time before the onset of the diet.

A controlled trial has now been started with new patients entering the AS clinic. It is concluded that diet modification in AS justifies further evaluation.


The cross-tolerance hypothesis provides a model with which to study the etiopathogenesis of ankylosing spondylitis. Refutable theories, starting as conjectures which can be disproved, are necessary tools in carrying out scientific investigations (75).

Erythrocyte sedimentation rate before and after 9 months on high protein-low carbohydrate regime in 36 active ankylosing patiients

Figure 12. Erythrocyte sedimentation rate before and after 9 months on high protein-low carbohydrate regime in 36 active ankylosing patiients.

The following statements, derived from the cross-tolerance hypothesis, are proposed as theoretical guidelines in carrying out further experimental and clinical studies:

1. HLA-B27 is directly involved in producing AS.
2. Gram-negative bacteria resemble HLA-B27.
3. The cross-tolerance hypothesis suggests that it is the antibacterial antibody acting as an autoantibody which triggers tissue damage and produces reactive arthritis.
4. AS is a form of reactive arthritis to gram-negative bacteria, and the lymphatic drainage of the colon and rectum define the anatomical features of the disease.
5. Several gram-negative bacteria can produce the disease, but in the urban London population subclinical gut infection by a commensal organism like Klebsiella is more likely to be a trigger factor than the more pathogenic species Salmonella, Shigella, Chlamydia. or Yersinia.
6. Substrate modulation through diet requires further evaluation in patients with AS.
7. Finally, the cross-tolerance hypothesis may have a wider application and could be relevant in the study of diseases other than ankylosing spondylitis.


The association of ankylosing spondylitis (AS) with HLA-B27 can be explained by a cross-tolerance hypothesis, in which it is suggested that the HLA-B27 gene product resembles common gram-negative antigens, and studies with tissue-typing sera and monoclonal anti-B27 antibodies are consistent with this hypothesis.
Reactive arthritis following Salmonella, Shigella, and Yersinia gut infections, occurs predominantly in HLA-B27-positive individuals, and Klebsiella microorganisms have been shown in some studies to occur more frequently in AS patients during active phases of the disease. Elevated levels of total serum IgA are clearly associated with biochemical parameters of disease activity, such as erythrocyte sedimentation rate and C-reactive protein levels. Studies using absorption and ELISA methods indicate that specific Klebsiella IgA antibodies can be demonstrated in patients with active AS.
It is suggested AS is a form of reactive arthritis following exposure to gram-negative bacteria and tissue damage is produced by antibacterial antibodies binding to cross-reactive self-antigens, such as HLA-B27, predominantly in the region of the lymphatic drainage of the colon and rectum, namely the lumbosacral area.

The cross-tolerance hypothesis implies that reduction of gram-negative bowel flora, by antibiotics or dietary means could be of therapeutic value in the treatment of AS. The cross-tolerance hypothesis as proposed for AS, resembles the etiopatho-genetic model proposed for rheumatic fever, whereby cross-reactive antigens on Streptococcus pyogenes evoke the formation of anticardiac, brain, and synovial antibodies, and the virtual elimination of this disease by the use of penicillin suggests that reduction of gram-negative antigenic load could reduce the severity of "B27 disease."


1. Aho, K. (1983): Yersinia reactive arthritis. Br. J. Rheumaiol.. 22(Suppl.2):41-45.
2. Aho, K., Ahvonen, P., Alkio, P., Lassus. A., Sairancn, E., Sievers, K., and Tiilikainen, A. (1975): HLA 27 in reactive arthritis following infection. Ann. Rheum. Dis., 34(Suppl.):29-30.
3. Aho, K., Ahvonen, P.. Lassus, A., Sievers, K., and Tiilikainen, A. (1973): HLA 27 in reactive arthritis. Lancet, ii:157.
4. Aho, K., Ahvonen, P., Lassus, A., Sievers. K., and Tiilikainen, A. (1974): HLA 27 in reactive arthritis: a study of Yersinia arthritis and Reiter's disease. Arthritis Rheum., 17:521-526.
5. Anderson, I. H., Levine, A. S., and Levitt. M. D. (1981): Incomplete absorption of the carbohydrate in all-purpose wheat flour. N. Engl. J. Mcd.. 304:891-892.
6. Arnett, F. C, Hochberg, M. C. Bias, W. B. (1977): Crossreactive HLA antigens in B27 negative Reiter's syndrome and sacroiliitis. Johns Hopkins Metl. I, 141:193-197.
7. Avakian, H., Welsh, J.. Ebringer, A., and Entwistle, C. C. (1980): Ankylosing spondylitis, HLA-