We found that SF levels of Δdi-C6S, Δdi-C4S, KS and the ratio of Δdi-C6S to Δdi-C4S (C6S/C4S) are significantly associated with the onset and progression of pre-radiographic high-grade focal cartilage damage—which was not reflected by a transition to detectable radiographic changes—based on longitudinal arthroscopic and radiographic evaluations. Many studies have reported on biomarkers associated with cartilage damage progression with regard to radiographic changes [18,19,20,21,22,23]. This study is the first to report that biomarkers of SF can predict pre-radiographic onset and progression of early cartilage damage on the basis of longitudinal arthroscopic observations.
In both the progression and non-progression groups, arthroscopic evaluations of morphological cartilage damage at baseline revealed similar numbers of grade 0–IV lesions (Table 2). In fact, morphologically normal cartilage and minimally damaged lesions (grade 0–I) exhibited onset and progression of cartilage damage, respectively, in the progression group (Table 2). Interestingly, cut-off values of 55.7 nmol/ml for Δdi-C6S, 10.6 μg/ml for KS, and 4.6 for C6S/C4S, as determined by ROC analysis (Fig. 2), correspond to grades 2–4 according to the Kellgren-Lawrence grading scale [6, 10, 24], or advanced OA [25]. This suggests that those showing progression of high-grade cartilage damage exhibited a cartilage aggrecan metabolism similar to that observed in patients with advanced radiographic OA [12], although no radiographic OA changes were observed in this study. Therefore, cut-off values of Δdi-C6S, KS, and C6S/C4S could be used as biomarkers to predict onset/progression of cartilage damage. Changes in proteoglycans, almost certainly released primarily from aggrecan because of its much greater content of these glycosaminoglycans, reflect cartilage metabolism at an early stage of OA [26]. Although the loss of aggrecan leads to increased aggrecan synthesis, newly synthesized molecules are composed of enriched C4S instead of C6S and KS [8,9,10], unlike those found in normal SF [16]. Our results are consistent with previous reports [8,9,10, 16], in that levels of Δdi-C6S and KS and the ratio of C6S/C4S were negatively correlated with cartilage damage progression.
With regard to Δdi-C4S, when this was added to the logistic regression analysis, a significant difference was noted such that progression of cartilage damage did not advance if Δdi-C4S was above 19.0 nmol/ml (calculated from the ROC curve) (Table 4), although a significant difference was not found between the progression group and non-progression group (Table 3), and the area under the ROC curve for Δdi-C4S was 0.618, which was lower than that for Δdi-C6S (0.746), KS (0.699), and C6S/C4S (0.689) (Fig. 2). Normally, Δdi-C4S is highly expressed in OA cartilage [8,9,10], which seems to contradict the result of this study. However, Δdi-C4S and Δdi-C6S showed a significant positive correlation (correlation coefficient = 0.838 [95% CI 0.744–0.899, p = 0]) with Pearson’s product-moment correlation in this study. We surmise that, with regard to early changes following trauma, as long as synthesis of Δdi-C6S was greater than that of Δdi-C4S, progression of cartilage damage could not advance. Therefore, it is appropriate to consider also the ratio of C6S/C4S as a predictive biomarker for progression of cartilage damage.
Our previous cross-sectional study found that C2C, which reflects increased collagenase cleavage of cartilage type II collagen, was significantly associated with the presence of arthroscopic high-grade cartilage damage [1]. C2C is considered a sensitive marker for detecting early cartilage damage. However, in the present longitudinal study, C2C was not associated with the onset/progression of high-grade cartilage damage. Mean SF C2C level in OA patients (n = 54) was reported to be 30.5 ng/ml in a previous study [27], whereas the median C2C level in the present study was 9.8 ng/ml (interquartile range, 5.8–13.8). Since our patients were at the pre-radiographic OA stage, their C2C levels might have been lower than those of OA patients and thus did not predict the progression of cartilage damage. In a recent study, a C2C-HUSA urine assay, which measures a more specific degradation product(s), revealed the association between baseline data and knee OA progression [18].
Older patients reportedly have increased cartilage damage and more severe OA changes after ACL injury [28]. Age is negatively correlated with the ratio of C6S/C4S [15], and the C6S/C4S ratio is typically lower in females than in males [15]. This is likely due to the decreased cartilage repair capacity in older patients, and may even reflect gender specificity. In a previous study, we found that both age and duration from injury to first surgery were positively correlated with the number of high-grade cartilage lesions [1]. Moreover, mean age and mean duration were higher in the progression group compared to the non-progression group (Table 1). In addition, baseline cartilage damage might be an important driver for cartilage damage progression. Thus, age, duration from injury to first surgery, sex, and the number of high-grade lesions (grades III and IV) at baseline should all be taken into account when predicting cartilage damage. As shown in Table 4, the levels of Δdi-C6S and KS and the ratio of C6S/C4S were associated with the progression of high-grade cartilage damage, even after adjusting for age, duration from injury to first surgery, sex, and the number of high-grade lesions (grades III and IV) at baseline in multivariable logistic regression analyses. Therefore, levels of Δdi-C6S and KS and the C6S/C4S ratio could serve as meaningful predictive biomarkers.
Meniscal tears or defects are risk factors for knee OA [29]. However, the proportion of high-grade meniscus damage in the progression group was similar to that in the non-progression group, at both the first and second evaluations (Table 1). The main component of the meniscus is type I collagen. In contrast to articular cartilage that mainly comprises type II collagen and contains abundant proteoglycans, the meniscus is highly deficient in the proteoglycans [30]. In this regard, meniscus damage is unlikely to contribute to differences in the biomarkers in both groups.
Whether ACL reconstruction performed at our institute is reproducible or not is of considerable importance. All ACL reconstructions were performed by three senior surgeons who had operated on more than 100 cases. Clinical results of ACL reconstruction and arthroscopic longitudinal changes of cartilage damage after ACL reconstruction in this study were comparable to those of previous reports [31, 32].
There are some limitations to this study. First, only a limited number of biomarkers were evaluated. Various biomarkers that reflect cartilage metabolism are available [33], including procollagen II C-propeptide (CPII) [34], which serves as an indicator of collagen synthesis. Indeed, the progression of OA might reflect collagen synthesis rather than collagen cleavage [35]. Thus, changes in collagen metabolism in the early stages of OA should be explored. Second, as this study was conducted retrospectively, the duration from the first to second evaluation varied by patient. That said, the duration was approximately 2 years in most patients in both groups, and given the lack of significant differences between the two groups, its influence was likely minimal. Third, the sample size is relatively small for evaluation of predictive biomarkers. Thus, a bigger group will be needed to demonstrate whether the obtained cut-off values are proper for prediction of cartilage damage progression. Finally, the present study examined cartilage damage after ACL injury based on longitudinal arthroscopic observation, but did not address whether cartilage damage will progress further to radiographic OA damage. A longer observation period would be required to address this issue.