Association of chemokine expression in anterior cruciate ligament deficient knee with patient characteristics: Implications for post-traumatic osteoarthritis

Abstract

Background:

Stromal cell-derived factor-1a (SDF-1α) and high mobility group box chromosomal protein (HMGB1) are chemokines that can drive post-traumatic osteoarthritis (PTOA) induced by anterior cruciate ligament (ACL) injury. However, the influence of patient characteristics on expression of those chemokines remains unclear. Our aim was to determine the relationship between chemokine expression in synovial fluid (SF) of the ACL deficient (ACL-D) knee and patient characteristics including time from injury, sex, and age.

Methods:

SF samples were collected immediately prior to the first-time ACL reconstruction (ACLR) from 82 patients. Expression of SDF-1α and HMGB1 was measured with human specific solid phase sandwich enzyme-linked immunosorbent assays. The expression levels between groups divided by time from injury, or age, or sex was compared using Student’s t-test. The association of SDF-1α or HMGB1 levels with those variables was determined using regression analysis and Pearson product-moment correlation coefficient.

Results:

Regression and correlation analysis indicated significant correlation between SDF-1α expression and time from injury in the cohort (r = −0.266, P=0.016, n = 82) and in females (r = −0.386, P=0.024, n = 34). Significant correlation was also observed between SDF-1α expression and age in the cohort (r = −0.224, P=0.043, n = 82) and in males (r = −0.289, P=0.046, n = 48). No significant correlation between HMGB1 expression and patient characteristics was detected.

Conclusions:

SDF-1α rather than HMGB1 might serve as a protein marker for monitoring the development of PTOA in the ACL-D knee, especially in female patients.

INTRODUCTION

Anterior cruciate ligament (ACL) injuries usually result from twisting and pivoting motions of the knee. Therefore, ACL tear frequently occurs in athletes and physically active. In clinic, ACL tear is rarely isolated but is usually associated with injuries to the menisci, knee cartilage, and even subchondral bone [1,2]. An ACL transection in vivo study in sheep demonstrated that meniscal loads significantly increased in ACL deficient (ACL-D) knees, which provided an explanation for the co-existence of ACL tear and meniscal injuries frequently observed in clinic [3]. Moreover, the joint instability caused by ACL tear may lead to cartilage degeneration, a hallmark of osteoarthritis (OA) [4]. Even with the reconstruction of damaged ACL, a significant amount of cartilage-damaging matrix metalloproteinases (MMPs) could still be upregulated in injured knees [5]. In a retrospective, cross-sectional study of patients with ACL injuries, the authors concluded that post-traumatic OA (PTOA) developed in ACL-D knees on average 10 years after injury and the pathological condition progressed with time [6].

Over the past two decades, a group of chemokines, such as stromal cell-derived factor-1a (SDF-1α), have been identified as crucial mediators of cartilage degradation in OA pathogenesis [7–9]. Chemokines are chemotactic cytokines, a group of small-sized proteins that are secreted by a wide range of cells and participate in inflammation response to regulate cell movement, differentiation and proliferation [10]. Studies have shown that SDF-1α, an 8-kDa chemokine, can be produced by bone marrow stromal cells and synoviocytes during OA pathogenesis [11,12]. In a Hartley guinea pig ACL transection model, the 12-month post-injury SDF-1α level was elevated by 166% in knee synovial fluid (SF) compared with that in normal controls. This level was positively correlated with the Mankin score of cartilage in ACL-D knees. Moreover, in patients with knee OA, SDF-1α level in SF was significantly positively correlated with radiographic severity assessed by the Kellgren–Lawrence grading scale [13,14].

In addition to SDF-1, another chemokine called high mobility group box chromosomal protein 1 (HMGB1) was recently discovered as a proinflammatory mediator in OA pathogenesis. HMGB1, previously known as HMG-1 or amphoterin, is a nuclear protein that can either be passively released from necrotic cells or actively secreted by cells involved in the inflammation process. For this reason, HMGB1 is listed as a member of damage-associated molecular patterns (DAMPs). In a bacteria-induced adjuvant arthritis rat model or in patients with rheumatoid arthritis (RA), significantly increased HMGB-1 expression was detected in the cytoplasm and extracellular matrix of the inflamed synovial tissue [15]. Our previous study also showed that HMGB-1 could synergize with proinflammtory cytokines or fibronectin fragments in upregulating the expression of cartilage-dissolving MMPs in articular chondrocytes [16]. Furthermore, a study by Schierbeck et al. showed elevated levels of HMGB1 in SF of patients with juvenile idiopathic arthritis (JIA) and suggested HMGB1 as an inflammation mediator in the pathogenesis of JIA [17].

However, the expression of SDF-1α and HMGB1 in human knees with ACL injury prior to the onset of PTOA has never been examined. How patient characteristics (time from injury, age, sex) influence the expression of those two chemokines still remains unclear. In the study reported in this paper, we hypothesized that expression of SDF-1α and HMGB-1 could be measured at protein level in human ACL-D knees prior to first-time ACL repair surgery and their expression was influenced by time from injury, age, and sex of the patients. Therefore, we aimed to determine the correlation between contents of SDF-1α or HMGB-1 in SF and time from injury, age, sex of the patients. The findings from our study will shed some light on the molecular mechanisms underlying the initiation and development of PTOA in the ACL-D knee.

METHODS

Patient characteristics

This study was approved by the University of Iowa Institutional Review Board (IRB) (No. 200702733) and all enrolled patients signed the informed consent form. Eighty-two patients (24.2 ± 10.2 years old) who were scheduled to receive a first-time ACL repair surgery were included in this study. In this cohort, based on whether time from injury was longer than 30 days, 57.3% patients were defined as acute injury and 42.7% as chronic injury. There were 48 males (30 with acute injury, 18 with chronic injury) who made up 61.5% of the cohort and 34 females (17 with acute injury and 17 with chronic injuries) who made up 38.5% of the cohort. According to age of the patients, 42.7% were teens and 57.3% were adults (Table 1).

Table 1.

Patient grouping by time from injury, age, and sex.

Time from injury (Days between injury and surgery); N (%)	Cohort N (%)		Teens (≤ 18 years old) N (%)		Adults (> 18 years old) N (%)
	Male	Female	Male	Female	Male	Female
≤ 30 Days (Acute); 47 (57.3%)	30	17	12	11	18	6
> 30 Days (Chronic); 35 (42.7%)	18	17	7	5	11	12
Total: 82 (100%)	48/82 (61.5%)	34/82 (38.5%)	19/82 (23.2%)	16/82 (19.5%)	29/82 (35.4%)	18/82 (21.9%)
			35/82 (42.7%)		47/82 (57.3%)

Acquisition of SF samples from ACL-D knees of patients with ACL injury

Eighty-two SF samples volume ranging from ~0.05 mL to ~6.5 mL were aspirated without lavage from 82 ACL-D knees immediately prior to ACL reconstruction surgery. Each SF specimen was transported from the operating room to the lab within 2 h of being collected. In order to remove cells or other particles, each sample was centrifuged at 1500 rpm for 20 min at 4°C. The supernatant was then aliqouted into sterile Nunc® CryoTube® vials (Nalge Nunc International, Rochester, NY, USA). Each vial was bar coded, inventoried, and stored at −80°C until further use.

Determination of SDF-1α expression levels in SF samples

The levels of SDF-1α in SF samples were quantitated with a highly sensitive human CXCL12/SDF-1α quantikine ELISA kit (R&D Systems, Minneapolis, MN, USA). This solid-phase ELISA assay employs a monoclonal antibody and an enzyme-linked polyclonal antibody specific for human SDF-1α to capture this chemokine in SF samples. The amount of SDF-1α is proportional to the color developed by an enzymatic reaction after a substrate solution was added to the SDF-1α immune complex. The optical density (OD) of each colored standard or sample was firstly read at 450 nm and then at 570 nm as the background OD. A linear regression standard curve was established (R² = 0.997) and the concentration of SDF-1α in each sample was computed.

Determination of HMGB1 expression levels in SF samples

Similar to the quantitative method for SDF-1α level determination in SF samples, the levels of HMGB1 were measured with a highly sensitive human HMGB1 enzyme-linked immunosorbent assay (ELISA) kit manufactured by Shino-Test Corporation (Kanagawa, Japan) and purchased from IBL International Gmbh (Hamburg, Germany). HMGB1 in each standard or sample can be specifically bound by purified anti-HMGB1 antibody coated in the wells of the assay strips. This immune complex is then recognized by peroxidase conjugated secondary antibody.

Statistical Analysis

The means and standard deviations of levels of SDF-1α, HMGB1, or total proteins in each group were calculated and compared with Student’s t-test (two-sample assuming unequal variance). A box and whisker plot was constructed to show the 1st and 3rd quantiles, the maximum and minimum, the median and mean and the outliers of the expression levels of SDF-1α or HMGB1 in groups by time from injury, age, or sex. Regression analysis and Pearson product-moment correlation coefficient were employed to assess the relationship between independent variables (time from injury, age, sex) and expression of SDF-1α or HMGB1. GraphPad Prism 8 statistics software (San Diego, CA, USA) and Microsoft Office 2016 Excel were used for statistical analysis. The difference was considered significant when P<0.05.

RESULTS

Higher expression of SDF-1α was observed in patients with acute injury while higher expression of HMGB-1 was detected in patients with chronic injury

The average level of SDF-1α in SF of patients with acute injury was 3240.39 ± 1339.45 pg/mL which was higher than that of patients with chronic injury (2853.33 ± 906.39 pg/mL) (P=0.061) (Figure 1(a)). The box and whisker plot showed that only two outliers on the high end in the acute group were identified while in the chronic group two on the high end and one on the low end were identified. Although the 1st quartile and the minimum levels of SDF-1α in the acute group were lower than those in the chronic group, the median, 3rd quartile, and maximum levels in the acute group were all higher than those in the chronic group (Supplementary Figure S1). However, the chronic group exhibited higher expression of HMGB1 than the acute group: 48.14 ± 76.63 ng/mL in the chronic group vs. 28.63 ± 18.93 ng/mL in the acute group. This difference in HMGB1 expression was not statistically significant (P=0.075) (Figure 1(b)). There were no outliers identified in the acute group while four outliers on the high end were identified in the chronic group. The median, 3rd quantile, and 1st quantile levels of HMGB1 in the chronic group were all slightly higher than those in the acute group (Supplementary Figure S1).

Figure 1.

Comparison of expression of stromal cell-derived factor-1a (SDF-1α) and high mobility group box chromosomal protein (HMGB1) in synovial fluid (SF) of the anterior cruciate ligament deficient (ACL-D) knee between acute and chronic groups divided by time from injury. The average levels of SDF-1α (a) and HMGB1 (b) in each group were plotted and compared. The difference was not statistically significant (P=0.061 for SDF-1α, P=0.075 for HMGB1). Error bars represent standard deviation.

Higher expression of HMGB1 and SDF-1α was detected in the teen patients than in the adult patients.

The average SDF-1α levels in teen and adult patients were 3193.12 ± 1037.59 pg/mL and 2986.55 ± 1285.49 pg/mL, respectively. This difference was not statistically significant (P=0.101) (Figure 2(a)). The box and whisker plot showed two outliers on the high end in the teen group, and one outlier on the high end in the adult group. The maximum, 3rd quantile, 1st quantile, and minimum levels of SDF-1α in the teen group were lower than those in the adult group (Supplementary Figure S1). The average HMGB1 level in teens was 49.30 ± 75.10 ng/mL, which was higher than that in adults (27.76 ± 22.21 ng/mL) (P=0.054) (Figure 2 (b)). The box and whisker plot showed three outliers in the teen group, and one outlier in the adult group, both on the high end. The 3rd quantile, median, 1st quantile, and minimum levels of HMGB1 in the teen group were all higher than those in the adult group (Supplementary Figure S2).

Figure 2.

Comparison of expression levels of stromal cell-derived factor-1a (SDF-1α) and high mobility group box chromosomal protein (HMGB1) in synovial fluid (SF) of the anterior cruciate ligament deficient (ACL-D) knee between teen and adult groups divided by age or sex of patients. The average levels of SDF-1α (a, c) and HMGB1 (b, d) in each group were plotted and compared. The difference in expression levels of SDF-1α or HMGB1 was not statistically significant between groups divided by age or sex (P=0.213 for SDF-1α between age groups, P=0.101 for SDF-1α between two sexes; P=0.054 for HMGB1 between age groups, P=0.387 for HMGB1 between the two sexes). Error bars represent standard deviation.

Female patients showed higher SDF-1α expression while male patients showed higher HMGB1 expression

The average level of SDF-1α in SF of ACL-D knees of female patients was 3279.90 ± 1283.54 pg/mL. This level was higher than that of male patients which was only 2930.17 ± 1098.03 pg/mL. However, this expression difference of SDF-1α was not statistically significant (P=0.101) (Figure 2(c)). The box and whisker plot showed three outliers on the high end, and one outlier on the low end in the female group while there were no outliers identified in the male group. The 3rd quantile, median, 1st quantile, and minimum levels of SDF-1α in the female group were all higher than those in the male group (Supplementary Figure S1). Compared to female patients who expressed 35.21 ± 24.49 ng/mL HMGB1, male patients expressed higher levels of HMGB1 (38.20 ± 65.84 ng/mL). But this difference in HMGB1 expression between two sexes was not statistically different (P=0.387) (Figure 2(d)). The box and whisker plot indicated that there were two outliers identified in each group, both on the high end. The 3rd quantile, median, 1st quantile, and minimum levels of HMGB1 in females were slightly higher than those in males (Supplementary Figure S2).

In the cohort, SDF-1α expression declined with time from injury and this negative association was statistically significant; however, no correlation was observed between SDF-1α and sex of the patients in the cohort

Concentrations of SDF-1α in SF of ACL-D knees of the cohort were plotted against time from injury. Pearson’s correlation analysis indicated that SDF-1α level was negatively correlated with time from injury of patients (r = −0.266). Regression analysis revealed that this correlation was statistically significant (P=0.016) (Figure 3(a)). By contrast, Pearson’s correlation coefficient between SDF-1α level and sex of the patients in the cohort was only 0.147 and regression analysis revealed that there was no correlation between those two variables (P=0.189) (Table 2).

Figure 3.

The relationship between stromal cell-derived factor-1a (SDF-1α) expression in synovial fluid (SF) of the anterior cruciate ligament deficient (ACL-D) knee and time from injury. (a) Scatter plot depicting the relationship between SDF-1α expression and time from injury in the cohort. (b) Scatter plot depicting the relationship between SDF-1α expression and time from injury in female patients. Linear regression analysis indicated significant correlation between those two variables in either the cohort (r = −0.266, P=0.016) or in female patients (r = −0.386, P=0.024).

Table 2.

Results of correlation and regression analysis on relation between chemokine expression and patient characteristics.

Independent Variables	Sex	SDF-1a			HMGB1
		Pearson correlation coefficient (r)	R2	P-value	Pearson correlation coefficient (r)	R2	P-value
Time from injury	Both sexes	−0.266	0.071	*0.016	0.047	0.002	0.673
	Female pts.	−0.386	0.149	*0.024	0.242	0.058	0.169
	Male pts.	−0.189	0.036	0.199	0.006	3.3 × 10−5	0.969
Age	Both sexes	−0.224	0.050	*0.043	−0.153	0.023	0.170
	Female pts.	−0.166	0.028	0.326	−0.261	0.068	0.137
	Male pts.	−0.289	0.084	*0.046	−0.145	0.021	0.326
Sex (Male = 1, Female = 2)		0.147	0.022	0.189	−0.028	0.001	0.801

^*P < 0.05

Significantly negative association between SDF-1α expression and time from injury was also detected in females but not in males

In female patients, SDF-1α expression in SF of ACL-D was also negatively correlated with time from injury. The Pearson’s correlation coefficient was −0.386 which indicated even stronger negative association between those two variables in females than in the cohort. Regression analysis revealed this correlation was also statistically significant (P=0.024) (Figure 3(b)). However, in male patients, Pearson’s correlation coefficient of those two variables was only −0.189. No correlation between SDF-1α and time from injury was detected (P=0.199) (Table 2).

Significantly negative association between SDF-1α expression and age was detected both in the cohort and in male patients

Between SDF-1α expression and age, those two variables in the cohort, Pearson’s correlation coefficient was −0.224. Regression analysis revealed that this correlation was statistically significant (P=0.043). This significantly negative correlation between SDF-1α expression and age was also detected in male patients (r = −0.289, P=0.046) (Figure 4(a), (b)). However, Pearson’s correlation coefficient was much lower in female patients and regression analysis revealed no correlation existing between SDF-1α and age (r = −0.166, P=0.326) (Table 2).

Figure 4.

The relationship between stromal cell-derived factor-1a (SDF-1α) expression in synovial fluid (SF) of the anterior cruciate ligament deficient (ACL-D) knee and age. (a) Scatter plot depicting the relationship between SDF-1α expression and age of the patients in the cohort. (b) Scatter plot depicting the relationship between SDF-1α expression and age of male patients. Linear regression analysis indicated significant correlation between those two variables in either the cohort (r = −0.224, P=0.043) or in female patients (r = −0.289, P=0.046).

Unlike SDF-1α, the expression of the other chemokine examined in this study, HMGB1, showed no correlation with any of those three variables in the cohort or in either sex

In the cohort, Pearson’s correlation coefficient between HMGB1 expression and time from injury, or age, or sex, was 0.047, −0.153, −0.028, respectively. The Pearson’s correlation coefficient between HMGB1 and time from injury or age in female patients was 0.242 and −0.261, respectively. In male patients, those values were even lower and changed to 0.006 and −0.145, respectively. Due to those low correlation coefficients, regression analysis revealed no correlation between those variables either in the cohort or in either sex (Table 2).

DISCUSSION

Our study revealed expression differences of two chemokines, SDF-1α and HMGB1, in the ACL-D knee SF prior to the onset of PTOA. Between two groups divided by time from injury, age, or sex of the patients, higher SDF-1α expression was detected in female or teen patients or in patients whose time from injury was shorter than 30 days, while higher HMGB1 expression was observed in male or teen patients or patients whose time from injury was longer than 30 days. Furthermore, regression analysis supported part of our hypothesis that SDF-1α expression was correlated with time from injury and age of the patients. To our knowledge, this is the first report that depicts the association between the expression of those two chemokines in SF and patient characteristics before PTOA develops in the ACL-D knee.

The severity of joint injury may positively correlate with the likelihood of PTOA. A study conducted by the biomechanics lab of our department showed that the possibility of the onset of PTOA in fractured ankle joint increased with the increase of the combined fracture severity scores generated from computed-tomography-based image analysis [18]. Although we did not assign injury severity scores to patients with ACL injuries in our study, based on the length of time from injury of the patients, we can reasonably assume that patients with a shorter time from injury may sustain more severe damage to the ACL, i.e. a higher injury severity score if being assigned. This led us to speculate that patients with a shorter time from injury may be more prone to developing PTOA in the injured knee than those with a longer time from injury.

Our data revealed that SDF-1α was negatively correlated with time from injury of the patients in the cohort. As a chemokine, the role of SDF-1α is to attract leukocytes to the inflammation sites [19]. In a PTOA guinea pig model and in a mouse model, both created by ACL transection (ACLT) surgical procedure, significantly elevated SDF-1α levels were observed in SF and serum, respectively, compared to normal or sham controls. This elevation was accompanied by the inflammatory responses, including upregulation of MMP-13 and interleukin-1b (IL-1b) in SF, severe cartilage damage assessed by the modified Mankin grading system, and increased bone resorption. Furthermore, inhibition of SDF-1 signaling in mouse with ACL transection by AMD3100 could reduce cartilage degeneration and attenuate PTOA development [14,20].

According to the number and spacing of the conserved cysteine (C) residues at the N-terminus of peptide chain, chemokines are divided into four types: C, CC, CXC, CXXXC. SDF-1α is a member of the CXC family which contains a motif composed of two cysteines separated by one intervening amino acid and was recently named CXCL12 [21]. It is exclusively expressed in higher vertebrates, including humans, and mainly attracts neutrophils [22]. The specific cell membrane receptor to SDF-1α/CXCL12 is C-X-C chemokine receptor type 4 (CXCR4) and the binding of SDF-1α/CXCL12 to CXCR4 activates intracellular signaling pathways related to chemotaxis, cell survival and/or proliferation, and calcium influx (22). For instance, the interaction between SDF-1α/CXCL12 and CXCR4 blocks the entry of HIV-1 into CD4(+) T cells [23].

Moreover, Kanbe and colleagues discovered that SDF-1α/CXCL12 was highly expressed in fibroblast-like synoviocytes isolated from synovium of osteoarthritic knee joints while the expression of SDF-1α/CXCL12 was barely detectable in the synovium of normal knees and was under-detected in chondrocytes isolated from cartilage of osteoarthritic knee joints. Conversely, the expression of CXCR4 was only detected in chondrocytes, not in fibroblast-like synoviocytes. Furthermore, they showed that cartilage-damaging MMP-3, −9, 13 could be upregulated in CXCR4-expressing chondrocytes by SDF-1α/CXCL12 secreted from fibroblast-like synoviocytes and synovectomy reduced the levels of SDF-1α/CXCL12, MMP-9 and −13 in serum of patients with OA [12,24]. In our study, SDF-1α expression in SF of the ACL-D knee increased with the shortening of time from injury. It implied that SDF-1α as an upstream pro-inflammatory might be heavily involved in the development of PTOA in the ACL-D knee by propelling the upregulation of cartilage-damaging MMPs.

We also discovered that the negative correlation between SDF-1α expression and time from injury was even stronger in female patients than in the cohort while this correlation was not observed in male patients. In addition, SDF-1α expression was higher in female patients than in male patients. Considering the role of SDF-1α in causing cartilage degradation, this gender difference echoed well with the gender dimorphism that has been reported for OA at epidemiology, radiographic and circulating biomarkers. A meta-analysis of sex differences in OA prevalence, incidence, and severity reported that females had significantly higher risk of prevalent OA in the knee and higher incidence in knee OA than males. The authors also discovered that females tended to have more severe radiographic knee OA than males [25].

Magnetic resonance imaging (MRI) comparison analysis indicated that women appeared to have lower femoral and patellar cartilage volumes than did men. Moreover, sex difference in patellar cartilage volume was magnified with age [26]. This may confer a more durable structural basis on men who have lower OA prevalence and incidence and less severe osteoarthritic symptoms than women. This speculation was supported by a biochemical biomarker study showing that the serum level of C-terminal telopeptide of collagen type I (CTX-I), a bone degradation marker, in men with OA was significantly lower than that in women with OA. This result implied that men with OA suffered less bone damage than women with OA [27].

Our study also revealed that significantly negative correlation between expression levels of SDF-1α in SF of the ACL-D knee and age existed in the cohort and in male patients. In other words, the younger a patient with ACL injury was, the higher level of SDF-1 was expressed in SF of the ACL-D knee, especially in male patients. This finding was consistent with what Bigoni et al. discovered in their study on levels of cytokines and chemokines in SF of teen patients (88% of them were males) with ACL tear. Their results showed that the expression of IL-8, a CXC chemokine like SDF-1α, was significantly higher in patients with ‘open’ physes than those with ‘closing’ or ‘closed’ physes. This suggested that stronger catabolic response to ACL injury might occur in teen patients (especially boys) than in adults [33].

HMGB1 as a ubiquitously expressed nuclear protein is responsible for chromatin stability in normal status. However, this nuclear protein can be released into the extracellular space during cellular stress or cell death. Besides this passive release, HMGB1 can be actively secreted by various types of cells, such as activated macrophages. Extracellular HMGB1 from passive release and active secretion can interact with receptors for advanced glycation end products (RAGE) and several components of the plasminogen activation system to mediate inflammation and tumor metastasis via upregulating tumor necrosis factor/IL-1 expression and enhancing the activity of tissue plasminogen activator and MMPs [28–30].

Although extracellular HMGB1 also acts as an inflammation mediator, we could not detect correlation between its expression and patient characteristics. Nonetheless, teen patients exhibited significantly higher levels of HMGB1 in SF of the ACL-D knee than did adult patients. This finding is consistent with the report of Iriuchishima et al. that the sagittal ACL area was negatively correlated with age in humans from studying the relation between age and the morphological characteristics of ACL with MRI [31]. Based on their discovery, teen patients have larger sagittal ACL area than adult patients, which may imply more cell death and higher levels of HMGB1 release from ruptured cell nuclei when the ACL is injured. Moderately higher HMGB1 expression was also observed in male patients than in female patients. Because ACL tears are frequently associated with meniscal, cartilage, and subchondral bone damage, dead cells in those damaged tissues can release significant amounts of HMGB1 into the SF. Given the gender difference in the histology of those joint tissues, it is not difficult to understand that men with ACL injury would release more HMGB1 than women with ACL injury.

We discovered that one of the two examined chemokines, SDF-1α, was negatively correlated with time from injury or age of the patients while the other mediator, HMGB1, did not show a clear correlation pattern. Those results implied that, compared to HMGB1, SDF-1α was more qualified as a prognostic protein marker for the onset of PTOA in ACL-D knees. This speculation is based on the observation that the injury severity of the ACL-D knee may positively correlate with the likelihood of PTOA while more severe ACL injury could lead to shorter time from injury. Our data showed that SDF-1α expression, not HMGB1 expression, in SF of the ACL-D knee was negatively correlated with time from injury. This might imply that SDF-1α expression could positively correlate with injury severity and then with the likelihood of PTOA in the ACL-D knee. Nonetheless, the role of SDF-1α as a PTOA prognostic biomarker needs to be validated with data of injury severity scores (ISS) generated from MRI images of ACL injured knees and osteoarthritic changes in ACL-D knees in our follow up studies.

Although studies have shown prevalence, incidence, and severity of knee OA are higher in women than men and around the time of menopause those differences are magnified even further [25,32], it still remains unclear whether women have a more accelerated rate of knee cartilage volume loss which is the hallmark of OA. The gender difference of SDF-1α in SF of the ACL-D knees identified in our study may shed some light on the molecular basis for the gender difference in incidence and severity of knee OA. Recent studies have linked SDF-1α to synovitis and cartilage degradation in OA. The effect of time from injury, age, and sex on SDF-1α expression in the ACL-D knee demonstrated in our study may aid us to understand the role of this chemokine in the development of ACL injury-induced PTOA in men and women in different age groups. Moreover, our discovery on gender and age difference in the expression of local proinflammatory mediators urges orthopedic surgeons to consider a specific anti-inflammation treatment regimen in women, especially teens, with ACL injury with the aim of preventing the early onset of PTOA in the ACL-D knee.

CONCLUSIONS

Through analyzing the levels of SDF-1α in SF of ACL-D knees, we discovered that, compared with male patients, female patients not only showed higher expression of SDF-1α but also exhibited an inverse relationship between SDF-1α expression and time from injury. Considering the role of SDF-1α in promoting inflammation in joints, this might offer an answer to the question of why women are more susceptible to knee OA than men. Conversely, we could not detect significant correlation of HMGB1expression with time from injury, or age, or sex of patients. Our work implied that SDF-1α rather than HMGB1 might be a potential protein marker for PTOA in the ACL-D knee, especially in female patients. The significance and innovations of our study reported in this manuscript are as follows. (1) Expression of SDF-1α and HMGB1, two chemokines promoting arthritic changes in weight-bearing joints, in SF of the ACL-D human knee measured before PTOA was developed in the affected knee. (2) Expression difference between groups divided by time from injury, age, and sex was compared and identified. (3) Correlation and regression analysis between independent variables (time from injury, age, and sex) and dependent variables (levels of SDF-1α and HMGB1 in SF of the ACL-D knee) were determined. The significant correlation between SDF-1α, not HMGB1, and time from injury or age of patients was detected. (4) Linking SDF-1α expression in SF of the ACL-D knee to injury severity score of the knee and to the likelihood of PTOA in the affected knee will shed some light on the molecular mechanism underlying the initiation and development of ACL-injury-induced knee PTOA.