The Natural Course of Arthrosis of the Knee

Abstract

The aim of this study was to describe the course of untreated or conservatively treated arthrosis of the knee joint. Of 265 patients presenting with knee pain who had weightbearing radiographs taken between 1970 and 1973, 132 (33 men and 99 women) responded to and participated in a clinical followup 20 years later using the Hospital for Special Surgery score and new weightbearing radiographs. Arthrosis was defined as an equal to or greater than 50% joint space narrowing. In 75 of 132 knees of Ahlbäck Class 0 (57%) and 20 of 52 knees of Ahlbäck Class I (39%), the classification remained unchanged. The Hospital for Special Surgery score at followup was compared with the Ahlbäck classes in the 1970s for the 110 cases for which surgery was not done. A higher Ahlbäck class at the time of presentation was associated with lower functional score (Hospital for Special Surgery) 20 years later. In addition, in the contralateral knee a reduction of joint space may occur. In the arthrosis group with total joint space reduction or attrition there was a clinical and radiographic deterioration that would merit surgical intervention to avoid unnecessary loss of function. However, in knees with equal to or greater than 50% joint space reduction (Ahlbäck I) and pain, a considerable number (39%) did not deteriorate radiographically during a 20-year period, and 25% remain free of pain. It appears that the long-term prognosis of milk knee arthrosis is not necessarily poor. A substantial number of these patients will not have progression of the arthrosis.

Danielsson and Hernborg⁵ described the morbidity of knee arthrosis defined as subchondral sclerosis in nonweightbearing radiographs. They found an increase with age, which later was confirmed by Anderson and Felson.² Hernborg and Nilsson⁸ described the natural course of untreated knee arthrosis in a study based on nonweightbearing radiographs of knees taken at the time of presentation 10 to 18 years earlier. At the followup, clinical examination and weightbearing radiographs showed that more than half had deteriorated clinically and radiographically. Women did worse than men. Spector et al¹⁸ described the natural course using the Kellgren-Lawrence classification¹⁰ and found that 33% deteriorated during an average of 11 years. Knees with pain at baseline did worse. Bauer,³ in a study of the natural history of gonarthrosis, stated that symptomatic femorotibial arthrosis must be treated surgically or it invariably would deteriorate.

Factors that may influence the natural course have been identified in the literature and include obesity,^2,13 meniscus injuries,¹² sport activities (with soccer being particularly dangerous),^11,16 and risk occupations.^6,14,19,20 The Framingham group found an increasing prevalence with age. Only 30% of the probands with joint space reduction had pain,⁶ as described earlier by Forman et al.⁷

The current study describes the natural course of untreated or conservatively treated arthrosis of the knee joint, defined as joint space narrowing as seen on weightbearing radiographs, in a 20-year prospective study.

PATIENTS AND METHODS

In the archives of the Department of Radiology in Malmö University Hospital, all radiographs have been saved since 1918. From 1970 to 1973 all radiographs (265 were found) of patients with knee pain and weightbearing radiographs of the knee joint were retrieved and evaluated. The evaluation of the radiographs from the time of inclusion and 20 years later consisted of measurements of joint spaces in millimeters, recording of cysts, osteophytes, and irregularities in tibial and femoral subchondral bone, and classification according to Ahlbäck¹ (1968) Classes 0 to III: Class 0 = less than 50% joint space narrowing (no arthrosis); Class I = equal to or greater than 50% joint space narrowing; Class II = 100% joint space narrowing; and Class III or more requires increasing attrition of the subchondral bone.

At followup 20 years later, 132 patients (33 men and 99 women) responded and participated in a clinical followup consisting of a Hospital for Special Surgery score⁹ evaluation and new weightbearing radiographs. Lost to followup were 38 patients who died and 95 who, because of senility or other illness, could not or would not participate(Fig 1). The mean age of the responders in 1970 to 1973 was 57 years (range, 36-73 years), and for those who did not respond or had died (133) the mean age was 59 years (range, 41-76 years). Fifty-six of the 133 (42%) nonresponders had arthrosis develop, which could be verified by the presence of their weightbearing radiographs (taken during the years before the followup) in the archives. Their radiographs also were analyzed, and their Ahlbäck classes in 1970 were as follows: 85 (64%) had Ahlbäck Class 0, 33 (25%) had Ahlbäck I, 11 (8%) had Ahlbäck II, 4 (3%) had Ahlbäck III or higher. Thus, in these respects, they did not differ from the responders.

Fig 1: Of the 265 patients who had weightbearing radiographs taken between 1970 and 1973, 110 were reexamined in 1990.

The Hospital for Special Surgery score⁹ contains subjective functions (pain, walking ability, and stair climbing) and objective functions (range of motion, extension lag, stability).

Excluding patients who, during the followup period, had surgical treatment other than diagnostic arthroscopy, left 110 to be included in the study. Of the patients who underwent invasive surgery (13 osteotomies and 10 total knee replacements), the knees that underwent operation were of Ahlbäck Class II or greater and were not included in the logistic regression analysis(Fig 1).

Statistical methods included chi squared test of the two modes of measurement and analysis of covariance of progressive change. Logistic regression was used to analyze the relative risks of change in the contralateral knee.

RESULTS

The mean duration of pain before the initial weightbearing radiograph was 2 years (range, 0-18 years). The mean age at presentation in 1970 was 57 years (range, 36-73 years) and at followup 76 years (range, 54-93 years). The mean followup time was 19 years (range, 18-20 years). Sixteen knees had meniscectomies before the first radiograph. Three of these had arthrosis develop in both compartments. Four knees had lateral arthrosis in 1970. In 1990, they all had arthrosis in the same compartment only. In the 1970s, the 132 responders had knee pain as their presenting symptom. One hundred five patients had mainly femorotibial and 27 had mainly patellofemoral pain.

The Hospital for Special Surgery score at followup was correlated with each of the Ahlbäck classes in the 1970s for the 110 cases that did not have surgical treatment (that is, no arthroplasty or no osteotomy). A higher Ahlbäck class at the time of presentation was associated with lower Hospital for Special Surgery scores 20 years later (Fig 2).

Fig 2: Graphs showing comparison between Hospital for Special Surgery score (HSS 1990) and Ahlbäck class (1970) in affected (A) and nonaffected (NA) knees.

Of the patients who had no radiologic signs of arthrosis in the 1970s, more than half (75 of 132) still had a normal radiologic appearance, and the rest had signs of arthrosis development, usually in the Ahlbäck Class I. In addition, 39% of those with Ahlbäck Class I remained unchanged. Approximately the same portion (42%) had moved to the Ahlbäck Class II. Of the index radiograms in Classes II and III, most cases deteriorated. Pain did not necessarily follow radiologic deterioration, except in the worst group; all Ahlbäck Class III patients got worse and had pain(Fig 3).

Fig 3: Diagram showing movement between Ahlbäck (A) classes during 20 years (joints). The number of painfree(pf) joints is attached to each subgroup.

Cartilage height (joint space) was measured in millimeters on the initial radiographs and at the time of followup. Women had a lower cartilage height than did men in their normal (defined as Ahlbäck Class 0 and pain free in 1990) knees. The average height of joint space (the mean cartilage height) from 1970 to 1973 was reduced in all knees during the next 20 years by an average of 1.2 mm joint space medially.

With regard to other radiographic signs of possible degeneration, such as osteophytes, irregular tibial and femoral surfaces, subchondral sclerosis, and patellar osteophytes, they were seen in small numbers, and a pattern could not be detected. Once arthrosis of the knee started to develop in men, the tendency to lose joint space was stronger than that seen in women(Table 1). A logistic regression analysis of relative risks to deteriorate in the contralateral knee was undertaken (only unilateral knee arthrosis included) if deterioration (1 mm reduction in joint space) had occurred in the affected knee. It was found that the risk to deteriorate was greater medially than laterally in Ahlbäck Class 0(Table 2). In cases with deteriorating arthrosis there was a high risk of pain (odd's ratio 3.9, confidence interval = 1.4-11.2).

TABLE 1: Rate of Arthrosis Deterioration in Patients in Relation to Gender

TABLE 2: Relative Risk of Deterioration (≥1 mm Joint Space Decrease) in Nonaffected Contralateral Knee^*

In the chi squared test of the two methods of measuring joint space, the Ahlbäck class and the millimeter measurement, the latter had a significantly higher detection capability of joint space reduction. A change in Ahlbäck class requires a reduction of 3 mm or more. When measured in millimeters, 73% of the knees deteriorated. In the Ahlbäck classification, 58% of the knees deteriorated (Table 3).

TABLE 3: Difference of Detection Capability of Joint Space Reduction^*

Using analysis of covariance it was found that an additional reduction of joint space by at least 0.3 to 0.4 mm in the nonaffected knee should be expected medially and laterally for each 1-mm reduction in the affected knee(Table 4). The influence of patient age on the development of arthrosis was significant. Those with early onset of arthrosis of the knee lost more joint space during the followup period than did those with onset later in life (Fig 4).

TABLE 4: The Change of Joint Space Narrowing in the Contralateral Knee^*

Fig 4: Graph showing the influence of patient age at onset of reduction of joint space after 20 years. Patient averages are given in 2-year age intervals.

DISCUSSION

In accordance with Hernborg and Nilsson,⁸ the results of the current study show that more than 70% of the knees with Ahlbäck Class I or higher deteriorate during a 20-year period. In the knees with milder arthrosis, Ahlbäck Class I, 39% remained unchanged(Fig 3). In the current study, in the nonaffected contralateral knees, a decrease in mean cartilage height during 20 years was observed (Table 4). This also was reported by Dacre et al⁴ when observing joint space in normal knees. Early onset is an important factor affecting the loss of cartilage height during 20 years (Fig 4). Arthrosis of the knee can be a bilateral disease with the contralateral knee affected by incipient arthrosis(Table 2) undetected by the Ahlbäck grading but supported by the millimeter measurements (Table 3). In the future, it may be possible to detect early cartilage changes through degeneration markers in the synovial fluid and serum.^15,17

Knees classified as Albäck 0 or I may not deteriorate(Fig 3) and possibly should not be selected for surgery solely on the basis of the radiographic findings. However, a significant increase of pain directly related to joint space reduction could not be detected by a detailed analysis of the current measurements. Increased pain during the years could be multifactorial and may not be caused solely by progression of the arthrosis.

Indirect signs of degeneration, such as osteophytes, irregular femoral surfaces, subchondral sclerosis, and patellar osteophytes, do increase in frequency during the years but in such small numbers that they cannot be considered reliable signs of arthrosis.

A possible drawback of the current study is that scoring according to the Hospital for Special Surgery score, which was introduced in 1983, was not included in the 1970s observations. Another possible drawback is that 10 knees that had a knee arthroplasty could represent a small selection bias. There also was some joint space reduction in the initially nonaffected knees(Table 2). The loss of joint space in this group was of a much lower magnitude and affected mainly the medial compartment.

Bauer³ stated that all knees with arthrosis will, if not treated, invariably deteriorate with time. However, according to the data of the current study, if early surgical intervention for mild arthrosis of the knee (Ahlbäck Class I) was performed, as many as 39% of the joints would have been treated unnecessarily.

The natural course of arthrosis of the knee should be the standard for evaluation of long term followup studies of knee arthroplasties. The development of arthrosis in the initially nonaffected knee should, according to the authors' analysis, be considered. In the arthrosis group with total joint space reduction or attrition, a clinical and radiographic deterioration was seen that would merit surgical intervention to avoid unnecessary pain and loss of function.