1131

THE LANCET

Pathogenesis of OsteoarthrosisIT is conventional to think of osteoarthrosis as a

disease entity. But this may be wrong, for currentwork suggests that the process may be analogous toheart-failure. Thus, " joint-failure " (osteoarth-rosis) can be brought about either by an increasein the functional demands on essentially healthytissues (so that joint-failure secondary to acetabulardysplasia might be viewed as analogous to heart-failure secondary to hypertension) or by a deteriora-tion in the functional capacity of the tissues them-selves-here joint failure after femoral-head infarc-tion would be analogous to heart-failure after myo-cardial infarction. Often the pathological processresponsible for the increased functional demand onthe joint or for the deterioration in the functionalcapacity of the joint tissues is obvious: in suchsituations osteoarthrosis is known as secondary.This term would be better dropped, both because itis redundant and because it raises by implication theopposing and even more misleading term of primaryosteoarthrosis. By primary osteoarthrosis is meantthose forms of the process in which the initiatingcause is unknown : idiopathic would therefore be abetter term.With respect to the pathogenesis of idiopathic

osteoarthrosis, three schools of thought exist. Someworkers believe that the initial abnormality may liein the osseous vasculature, some that it lies in thebone itself, and some that it lies in the articular

cartilage. Probably the vascular school has fewadherents these days. The belief that vascularfactors are responsible for the disorder rests on

the demonstration by phlebography of abnormalvenous drainage from the bones in an osteoarthrosicjoint. 1-3 Although the abnormality does exist,probably it is secondary to other features of theosteoarthrosic process and not the cause of them.4 4

Venous obstruction in man does not cause osteo-arthrosis.There is no doubt that gross abnormalities of the

bony architecture-for example, traumatic deforma-tion of the joint surface-can cause osteoarthrosis.But in idiopathic osteoarthrosis no such gross changescan be demonstrated early in the disease. RADINet al., however, have suggested that bony changes1. Meriel, P., Ruffie, R., Fournie, A. Revue Rheumat. 1955, 22, 238.2. Brookes, M., Helal, B. J. Bone Jt Surg. 1968, 50B, 493.3. Hulth, A., Hernborg, J. ibid. p. 227.4. Freeman, M. A. R. in Modern Trends in Orthopædics (edited by

A. Graham Apley); p. 40. London, 1972.

of a much more subtle and clinically undetectablenature may be at work: they have suggested that theremay be an increase in the rigidity of subchondralcancellous bone which so impairs the joint’s capacityto damp peak dynamic stresses as to damage thecartilage.5 They suggest that increased rigidity maybe caused by callus formation secondary to isolatedtrabecular fatigue fractures (fractures of this kindcertainly do occur in the femoral head 6). In con-

formity with this general view, Foss and BYERS havedemonstrated that osteoarthrosis is more frequentin subjects with an above-average skeletal mass (andhence, by inference, with more rigid bones).

Although it is possible that factors of this kindmay occasionally be responsible for osteoarthrosis,most workers agree with COLLINS 8 that the firsttissue to be affected is cartilage, in which the synovialsurface undergoes fragmentation-a change knownas fibrillation. (Confusingly, fibrillation seems to

exist in two forms-distinguishable from each otherin some, but not in all, joints by their position,pathogenesis, and prognosis-one form being benignand age-related, the other pre-osteoarthrosic.9)Fragmentation of the surface must involve frag-mentation of the collagen network in cartilage, and,both histologically and biochemically,lo-13 fibrilla-tion is associated with depletion of the other majorstructural component of cartilage, proteoglycan.

Cartilage carries load by virtue of the interactionof its collagen, proteoglycan, and bound water 14:the proteoglycan retains the water and the collagenprobably retains the proteoglycan.15 Thus, primarydamage to the collagen might lead to proteoglycanloss by leakage or as a consequence of cellular injury.Alternatively, in so far as proteoglycan and watersupport the collagen, a primary loss of proteoglycanmight lead to overstressing, and hence fracture

(fragmentation), of the collagen. Which, then,comes first-proteoglycan depletion or collagenfracture ? Since there is no reason to suppose that

idiopathic osteoarthrosis is a single disease entity,these are not mutually exclusive alternatives. Histo-

logically the issue could be decided if it were possibleconvincingly to demonstrate either surface frag-mentation in the absence of proteoglycan depletionor proteoglycan depletion in the absence of surfacefragmentation. The former can be demonstrated

5. Radin, E. L., Paul, I. L., Tolkoff, M. J. Arthr. Rheum. 1970, 13, 400.6. Todd, R. C., Freeman, M. A. R., Pirie, C. J. J. Bone Jt Surg.

1972, 54B, 723.7. Foss, M. V. L., Byers, P. D. Ann. rheum. Dis. 1972, 31, 259.8. Collins, D. H. in The Pathology of Articular and Spinal Diseases;

p. 74. London, 1949.9. Byers, P. D., Contepomi, C. A., Farkas, T. A. Ann. rheum. Dis.

1970, 29, 15.10. Matthews, B. F. Br. med. J. 1953, ii, 660.11. Bollet, A. J., Handy, J. R., Sturgill, B. C. J. clin. Invest. 1963,

42, 853.12. Bollet, A. J., Nance, J. L. ibid. 1966, 45, 1170.13. Mankin, H. J., Lippiello, L. J. Bone Jt Surg. 1970, 52A, 424.14. Adult Articular Cartilage. Edited by M. A. R. FREEMAN. London:

Pitman Medical. New York: Grune and Stratton. 1973. Pp. 356.£9.50; $25.

15. Freeman, M. A. R., Kempson, G. E. ibid. p. 228.

1132

occasionally, but such proteoglycan depletion is

patchy and is confined to the surface layer 16: itsrelevance is uncertain. Certainly mechanical softening(which occurs when cartilage is depleted of proteo-glycan) is seen only when the surface is fibrillated.17,18Surface fragmentation can be seen with the electronmicroscope in the absence of proteoglycan deple-tion,19 but the borderline between surface normalityand abnormality is hard to define at the ultra-structural level. Thus, histological studies provideno convincing grounds for deciding which ab-

normality is primary; but such evidence as there isis marginally in favour of collagen fragmentation.

If proteoglycan depletion is primary (i.e., if it isnot due to " leakage " of proteoglycan through afragmented collagen net nor cellular damagefollowing fragmentation) it might be caused eitherby a failure of synthesis, excessive degradation, orthe synthesis of abnormal proteoglycan (for example,of proteoglycan complexes of an abnormally smallhydrodynamic size so that they can

" leak " from anormal collagen mesh). With regard to synthesis,investigations based on the incorporation of 35S

(as S04) appear to show not merely that the syn-thesis-rate is maintained but that it is increased infibrillated cartilage. 20-22 Although the conclusionto be drawn is debatable,23,24 conceivably the cellsare attempting, but failing, to maintain a normal

proteoglycan concentration in the matrix in the faceof the loss of this material from the matrix. With

regard to excessive depletion, lyzosomal enzymescapable of degrading proteoglycan (specificallycathepsin D and Bi) have been demonstrated in

cartilage 25 and such enzymes soften cartilage invitro .26 As yet there is no evidence that these

enzymes are unusually active in fibrillated tissue.There is some evidence for an abnormality in themolecular structure of the proteoglycans in fibrillatedand unfibrillated dog cartilage, but the mechanicaland pathological implications of this abnormalityare not clear.27 Thus, there is little concrete evidenceto support the notion that proteoglycan depletion,although it occurs when cartilage fibrillates, is the

primary event in idiopathic osteoarthrosis. Even ifthis were to be the case, the ultimate cause of osteo-arthrosis would still not be revealed, for the questionswould remain, What causes the biochemical ab-

16. Ghadially, F. N., Meachim, G., Collins, D. H. Ann. rheum. Dis.1965, 24, 136.

17. Kempson, G. E., Maroudas, A. Unpublished.18. Kempson, G. E. in Adult Articular Cartilage (edited by M. A. R.

Freeman); p. 198. London, 1973.19. Meachim, G., Roy, S. J. Bone Jt Surg. 1969, 51B, 529.20. McElligott, T. F., Collins, D. H. Ann. rheum. Dis. 1960, 19, 31.21. Collins, D. H., McElligott, T. F. ibid. p. 318.22. Mankin, H. J., Lippiello, L. J. Bone Jt Surg. 1970, 52A, 424.23. Stockwell, R. A., Meachim, G. in Adult Articular Cartilage (edited

by M. A. R. Freeman); p. 89. London, 1973.24. Maroudas, A. Biochim. biophys. Acta (in the press).25. Dingle, J. T., Barrett, A. J., Weston, P. D. Biochem. J. 1971,

123, 1.26. Kempson, G. E., Dingle, J. T., Barrett, A. J. Cited in Adult

Articular Cartilage (edited by M. A. R. Freeman); p. 196. London,1973.

27. Muir, H. Personal communication.

normality, and why is this abnormality induced

(if it is) in secondary osteoarthrosis due to apparentlymechanical factors such as acetabular dysplasia ?Presumably mechanical damage to the cells may bethe answer.

Turning from the rather unpromising argumentsfor proteoglycan depletion as being primary, someof the newer work makes a primary failure in thecollagen network look possible. The proposedmechanism for the failure of the surface, and laterdeep, collagen network is fatigue, probably leadingto the separation of individual fibres rather thantheir fracture. Bone, a connective tissue in manyways similar to cartilage, save that it is calcified, isprone to fatigue in the laboratory,28 and fatigue-fractures in life have been recognised both micro-scopically and grossly.6,29 Cartilage has also beenshown to be fatigue-prone in the laboratory, andanatomically the appearance of the surface of cartilagewhich has undergone fatigue failure by cyclicalcompressive loading in vitro resembles that offibrillated cartilage. 30 The stresses required to

produce these changes in the laboratory are higherthan those thought to act in normal joints but nothigher than the probable stresses in joints displayingincongruity of the sort recognised clinically as pre-osteoarthrosic. Thus, this mechanism could easilybe responsible for fibrillation leading to secondaryosteoarthrosis. That a similar mechanism may be atwork in idiopathic osteoarthrosis at the hip is sug-gested by the finding that in certain apparentlynormal hips from cadavers over the age of 50 thepeak contact pressure is far higher than the pressurespreviously supposed to exist-higher than thosefound in younger joints, and great enough to causefatigue failure in cartilage.31 The increase of contactpressures with age suggests that a disturbance ofbone remodelling may be at work. 32 Thus, in a

sense, bone is once again under suspicion as the seatof primary disease. High contact pressures in

apparently normal hips might be viewed as analogousto symptomless systemic hypertension, and perhapsthey provide the clue to the pathogenesis of manycases of idiopathic osteoarthrosis at the hip. If this

explanation holds and fibrillation in this conditionproves to be due to cartilage-collagen fatigue failure(with proteoglycan depletion as a secondary eventcaused by its " leakage " through the fragmentedcollagen network or its degradation by enzymesreleased from mechanically damaged cells), the

pathogenesis of primary and secondary osteo-

arthrosis would be unified. Finally, the thought,

arises that if the repetitive loads applied to articular

28. Swanson, S. A. V., Freeman, M. A. R., Day, W. H. Med. biol.Eng. 1971, 9, 23.

29. Morris, J. N., Blickenstaff, L. D. Fatigue Fractures. Springfield,1967.

30. Weightman, B. O., Freeman, M. A. R., Swanson, S. A. V. Nature,1973, 244, 303.

31. Freeman, M. A. R., Swanson, S. A. V., Day, W. H. Unpublished.32. Johnson, L. C. Lab. Invest. 1959, 8, 1223.

1133

cartilage are capable of producing fatigue in its col-lagen, and hence osteoarthrosis, could the repetitiveloads applied to the walls of arteries be responsiblefor some examples of degenerative arterial disease ?

Cancer and FoodTHE appearance of a cancer is preceded by circum-

stances which, in combination, can be called the" cause " of the cancer. Often there are a number of

necessary factors, none of which is sufficient on itsown. For instance, a specially inbred strain of micedevelops both breast and liver tumours in 100% ofanimals. We could say that there is a genetic causeunderlying these tumours. But move the strain ofmice from their home in Washington, away to

Adelaide, and now only 17% liver tumours and 10%breast tumours develop, until you bring both foodand sawdust bedding from the U.S.A., and then thecancers develop in the new immigrants to Australia,as they did before. 1 (The molecules responsibleremain unknown.) Occasionally an overwhelmingexposure to a chemical carcinogen produces a cancerin all those exposed. In the rubber industry, p-naphthylamine caused bladder cancer in all thosemen who had a large enough dose. But more oftenwe are working in circumstances where heredity andenvironment interact, and of particular concern isfood. Food contains low doses of carcinogen 2and food contains substances that alter metabolismand so alter the response to carcinogens.l,3 Wecannot as yet predict from animal work how thehealth of a human population will respond to exposureto a carcinogen. When one hears that a chemical

present in the environment causes cancers in animalsone’s thoughts oscillate between " Ban it at once "and " It’s a useful compound, pay no attention ".

Aflatoxins are metabolites produced by manystrains of Aspergillus flavus. The rnoulds can growin warm moist conditions on almost any vegetablefood. 4-6 Spores, moulds, or aflatoxins commonlycontaminate badly stored and harvested peanuts(groundnuts), maize, and cotton seed. Aflatoxin Biis the most potent known hepato-carcinogen and isactive for very many species, including trout, ferrets,and rats. 5, 7 Aflatoxins are often present in market

samples of food in tropical countries, especially inIndia and Africa, and aflatoxin metabolites havebeen found in the urine of children who had eaten

heavily contaminated home-made peanut butter in the

1. Sabine, J. R., Horton, B. J., Wicks, M. W. J. natn. Cancer Inst.1973, 50, 1237.

2. Crosby, N. T., Foreman, J. K., Palframan, J. F., Sawyer, R.Nature, 1972, 238, 342.

3. Wattenberg, L. W., Page, M. A., Leong, J. L. Cancer Res. 1968,28, 934.

4. Mehan, V. K., Chohan, J. S. Mycopath. Mycol. appl. 1973, 49,263.

5. Aflatoxin: Scientific Background, Control, and Implications(edited by L. A. Goldblatt). London, 1969.

6. Lopez, A., Crawford, M. A. Lancet, 1967, ii, 1351.7. Barnes, J. M., Butler, W. H. Nature, 1964, 202, 1016.

Philippines, Naturally, one is inclined to associateaflatoxin with the high incidence of liver cancer inmany tropical countries.9 "* Surveys in Swaziland andPapua suggest that more aflatoxin in the foodstuffsgoes together with more liver cancer in the population.Now an investigation from Kenya turns this impres-sion into a quantitative association. PEERS and

LINSELL, working for the Tropical Products Institutein London and the W.H.O. International Agencyfor Research in Cancer, have conducted two large-scale surveys in the Murang’a district of Kenya,which has an area of about 700 sq. miles, and apopulation of over 300,000, almost entirely Kikuyu.Liver-cancer incidence was measured by followingcases from the four hospitals, three health centres,and twenty-five dispensaries. A high proportion ofhistological confirmations was achieved, togetherwith measurement of alpha-fetoprotein. Studies oftuberculosis incidence and incidence of cancers ofsites other than the liver suggest that few cases ofliver cancer were missed, and the incidence-rateswere accurate. Total cancer incidence was 19-9 per100,000 per year over four years, while liver cancercontributed 10.8 and 3-8 cases per 100,000 per yearin men and women, respectively. Dietary investiga-tions involved measurement of aflatoxins in 2400

samples of prepared food in a ready-to-eat state,

gathered over two years in a properly distributedsampling scheme. The liver-cancer incidence invarious areas varied from nil to 12.9 cases per100,000 per year, while aflatoxin intake varied from3-4 to 14.8 ng. per kg. body-weight per day. A near-linear relation between liver cancer incidence and

log aflatoxin intake fits the data observed.This is by far the best study of aflatoxin intake

and liver-cancer incidence so far published. Itcombines excellent sampling of foods and beers,good techniques for aflatoxin measurement, and

convincing measures to find the true liver-cancerincidence in the populations studied. The con-clusions can only be tentative since there are othervariables (e.g., altitude and rainfall) apart fromaflatoxin between the areas of high and low liver-cancer incidence. But this work goes a long way toturn a clinical impression into a measurable, assess-able relation. One of the most important aspects ofthe investigation lies in the possibility of settingrealistic limits to

" permissible " aflatoxin con-

tamination. It is no use condemning foodstuffs asunfit for human use because of aflatoxin, if reducedfood supplies lead to deaths from starvation. The

finding of no liver cancer at an intake of 3 ng. per kg.body-weight per day of aflatoxin Bl (0-1 1 .g. per kg.diet, or 0-1 p.p.b.) gives a level to aim at. This ismuch better than aiming at zero aflatoxin, because" zero" depends only on the sensitivity of the

8. Campbell, T. C., Caedo, J. P., Bulatao Jayme, J., Salamat, L.,Engel, R. W. ibid. 1970, 227, 403.

9. Peers, F. G., Linsell, C. A. Br. J. Cancer, 1973, 27, 473.