An inflammatory PBL gene expression predicts fast radiographic progressors in both the NYU and OAI cohort. We recruited a total of 243 SKOA patients, followed in the clinics of NYU and analyzed the PBL inflammatory gene expression as a predictive biomarker of radiographic progression. The baseline demographic and clinical characteristics of these subjects are summarized in Table 1. Of the 243 patients in the NYU cohort, 30 percent exhibited ≥0.5 mm mJSN at 24 months in the signal knee and were designated “fast-progressors” (8, 10, 42). PBL mRNA expression of IL-1β, COX-2 and TNFα, at baseline significantly predicted fast radiographic progressors with AUCs that ranged from 0.62 to 0.68 (p = 0.003 to <0.0001) (Table 2).
Table 1. NYU and OAI cohorts SKOA subject baseline demographics and imaging features.
Variable
|
NYU (n=243)
|
OAI (n=203)
|
Age (years)
|
60.1 (10.6)
|
62.8 (10.6)
|
Sex (%):
|
|
|
Male
|
33.30%
|
47.0%
|
Female
|
66.70%
|
53.0%
|
BMI (kg/m2)
|
26.7 (3.6)
|
27.3 (3.1)
|
VAS (0-100)
|
42.3 (29.9)
|
NA
|
WOMAC (0-100)
|
36.6 (24.7)
|
18.5 (16.2)
|
Radiographic Joint space width- JSW (mm):
|
|
|
Baseline
|
3.65 (1.34)
|
3.62 (1.30)
|
24 months
|
3.13 (1.51)
|
3.36 (1.49)
|
Radiographic Osteophytes:
|
|
|
Medial tibial plateau (MTP) (0-3)
|
0.50 (0.70)
|
1.08 (1.17)
|
Medial femoral condyle (MFC) (0-3)
|
0.87 (0.69)
|
0.93 (1.02)
|
Lateral tibial plateau (LTP) (0-3)
|
0.41 (0.64)
|
0.99 (0.80)
|
Medial femoral condyle (LFC) (0-3)
|
0.51 (0.60)
|
0.33 (0.38)
|
MRI:
|
(n=111)
|
(n=46)
|
Mean medial BML WORMS score (0-15)
|
1.13 (1.86)
|
1.90 (2.04)
|
Data shown are the mean (SD), total number or percent affected, as indicated. SKOA=symptomatic knee OA; BMI=body mass index; VAS=Visual Analog Scale; WOMAC=Western Ontario and McMaster Universities Osteoarthritis Index; WORMS=Whole-organ Magnetic Resonance Imaging Score; BML=Bone Marrow Lesion; NA- not available. Both medial and lateral osteophytes in tibial plateau (MTP, LTP) and femoral condyle (MFC, LFC) regions scored semi-quantitatively (0-3) [0= absent, 1-mild, 2-moderate, and 3= severe].Subarticular bone marrow lesions (BML, a score of 0-3) were scored within the anterior, central, and posterior regions of the medial and lateral tibial plateaus; central and posterior regions of the medial and lateral femoral condyles were also scored.
Table 2. The area under the receiver operating characteristic curves (AUC) of inflammatory PBL gene expression biomarkers to radiographic fast progressors of JSN (>0.5mm) at 24 months in patients with SKOA in NYU extended and OAI cohorts.
Fast Progressors (JSN>0.5mm) vs. Non-Progressors
(JSN <=0.0mm)
|
AUC
(95% CI)
|
Adjusted P value for model
|
NYU (n=243)
|
Age
|
0.48 (0.39-0.56)
|
0.711
|
Sex
|
0.62 (0.54-0.71)
|
0.001
|
BMI
|
0.61 (0.53-0.69)
|
0.004
|
Age + Sex + BMI (ASB)
|
0.65 (0.57 - 0.73)
|
<0.0001
|
IL-1β
|
0.62 (0.53 - 0.70)
|
0.003
|
COX-2
|
0.68 (0.60 - 0.75)
|
<0.0001
|
TNFα
|
0.66 (0.57 - 0.74)
|
<0.0001
|
IL-1β + COX-2 + TNFα
|
0.60 (0.5 2- 0.69)
|
0.008
|
IL-1β + ASB
|
0.65 (0.57 - 0.73)
|
<0.0001
|
COX-2 + ASB
|
0.66 (0.58 - 0.74)
|
<0.0001
|
TNFα + ASB
|
0.64 (0.56 - 0.72)
|
<0.0001
|
IL-1β + COX-2 + TNFα + ASB
|
0.66 (0.58 - 0.74)
|
<0.0001
|
OAI (n=203)
|
Age
|
0.38 (0.27-0.49)
|
0.987
|
Sex
|
0.50 (0.39-0.61)
|
0.482
|
BMI
|
0.38 (0.27-0.48)
|
0.989
|
Age + Sex + BMI (ASB)
|
0.36 (0.26 - 0.47)
|
0.995
|
IL-1β
|
0.76 (0.66 - 0.85)
|
<0.0001
|
COX-2
|
0.64 (0.53 - 0.75)
|
0.006
|
TNFα
|
0.68 (0.58 - 0.79)
|
<0.0001
|
IL-1β + COX-2 + TNFα
|
0.69 (0.59 - 0.79)
|
<0.0001
|
IL-1β + ASB
|
0.63 (0.53 - 0.73)
|
0.007
|
COX-2 + ASB
|
0.50 (0.38 - 0.61)
|
0.513
|
TNFα + ASB
|
0.56 (0.45 - 0.67)
|
0.158
|
IL-1β + COX-2 + TNFα + ASB
|
0.59 (0.48 - 0.69)
|
0.063
|
NYU + OAI combined (n=436)
|
Age
|
0.47 (0.40-0.54)
|
0.822
|
Sex
|
0.56 (0.49-0.63)
|
0.039
|
BMI
|
0.57 (0.50-0.64)
|
0.021
|
Age + Sex + BMI (ASB)
|
0.57 (0.50 - 0.63)
|
0.019
|
IL-1β
|
0.66 (0.60 - 0.72)
|
<0.0001
|
COX-2
|
0.66 (0.59 - 0.72)
|
<0.0001
|
TNFα
|
0.67 (0.60 - 0.73)
|
<0.0001
|
IL-1β + COX-2 + TNFα
|
0.63 (0.57 - 0.70)
|
<0.0001
|
IL-1β + ASB
|
0.60 (0.54 - 0.67)
|
0.001
|
COX-2 + ASB
|
0.59 (0.53 - 0.66)
|
0.003
|
TNFα + ASB
|
0.57 (0.50 - 0.64)
|
0.019
|
IL-1β + COX-2 + TNFα + ASB
|
0.61 (0.54 - 0.68)
|
0.001
|
ASB= Age, Sex and BMI; JSN=joint space narrowing; SKOA=symptomatic knee OA; AUC=Area under the (receiver operating characteristic) curve; 95% CI= 95% confidence interval; All comparisons are versus non-progressors (JSN ≤0 mm). NYU cohort: non-progressors (n=115) and fast progressors (n=72); OAI cohort: non-progressors (n=76) and fast progressors (n=44).
To replicate these NYU Cohort PBL transcriptomic studies in an independent population, we next examined radiographic progression in an OAI Cohort (n=204) of subjects with symptomatic knee osteoarthritis. “Fast progression” was observed in 22 percent of the OAI Cohort. In the NYU cohort, age as covariant did not have significant predictive power, whereas sex and BMI did have significant but moderate AUC values 0.62 and 0.61 (p=0.001 and 0.004) (Table 2) in predicting fast progressors. In contrast, in the OAI cohort these variants did not achieve significance. We note that while age and BMI were similar in both cohorts, the OAI population had a higher percentage of females (NYU: 33%; OA: 47%). As shown in the combined cohort, both sex and BMI retained significance but with moderate predictive power in predicting fast progressors. Increased mRNA expression of IL-1β, COX-2 and TNFα, (levels between non-progressors and progressors are presented in supplemental Table 1) significantly predicted JSN fast progressors (≥0.5 mm at two years) with AUCs that ranged from 0.64 to 0.76 (p = 0.006 to <0.0001) (Table 2). In the combined NYU and OAI cohort of 447 subjects, the significance of an inflammatory PBL gene expression was retained (Table 2) in predicting fast radiographic progressors (AUC 0.66 to 0.67 (p <0.0001).
MRI cross-sectional imaging, radiographic, and symptom relationships. The association of baseline MRI-scored variables with clinical and radiographic features at baseline and 2 years is shown in Table 3. Baseline medial tibial central BML scores associated significantly, but modestly, with baseline WOMAC scores for pain (r=0.19, p=0.048) and 24-month VAS pain reports (r=0.20, p=0.043) (Table 3). Additionally, medial tibial central BML associated moderately with baseline KL scores and associated inversely with baseline JSW (Table 3, r=0.21, p<0.01; r= -0.22, p=0.018, respectively). Cartilage scores in all WORMS medial subregions, separately and/or summed, were also associated with baseline KL scores (r values ranging from 0.27 to 0.34 depending on the specific subregion, all p values <0.01; Table 3). Similarly, cartilage scores in nearly all medial subregions were inversely associated with baseline JSW (r values ranging from -0.27 to -0.33, all p values <0.01). Meniscus overall readings were also associated with baseline WOMAC pain and associated inversely with mJSW at 24 months (r=-0.34, p<0.01).
Table 3. Association of baseline MRI-scored medial WORMS variables with clinical and radiographic features at baseline and 24 months – NYU Cohort (n=111).
|
Baseline WOMAC pain
|
Baseline VAS pain
|
24 Month WOMAC pain
|
24 Month VAS pain
|
Baseline Signal
KL
|
24 Month Signal KL
|
Delta Signal KL
|
Baseline Medial JSW
|
24 Month Medial JSW
|
Medial JSN
|
Cartilage Overall
|
0.09 (0.34)
|
0.01 (0.94)
|
0.07 (0.45)
|
0.10 (0.29)
|
0.33 (<0.01)
|
0.44 (<0.01)
|
0.20 (0.03)
|
-0.28 (<0.01)
|
-0.34 (<0.01)
|
0.13 (0.15)
|
Cartilage Medial Overall
|
0.15 (0.12)
|
0.12 (0.22)
|
0.09 (0.34)
|
0.09 (0.38)
|
0.36 (<0.01)
|
0.44 (<0.01)
|
0.14 (0.15)
|
-0.32 (<0.01)
|
-0.41 (<0.01)
|
0.10 (0.29)
|
BML Overall
|
0.07 (0.45)
|
0.09 (0.36)
|
0.04 (0.66)
|
0.10 (0.28)
|
0.31 (<0.01)
|
0.35 (<0.01)
|
0.05 (0.58)
|
-0.13 (0.16)
|
-0.22 (0.02)
|
0.14 (0.14)
|
BML Medial Overall
|
0.10 (0.30)
|
0.13 (0.18)
|
0.04 (0.69)
|
0.07 (0.48)
|
0.31 (<0.01)
|
0.32 (<0.01)
|
-0.003
(0.98)
|
-0.17 (0.07)
|
-0.28 (<0.01)
|
0.19 (0.05)
|
Meniscus Overall
|
0.20 (0.04)
|
0.15 (0.11)
|
0.08 (0.38)
|
0.13 (0.18)
|
0.30 (<0.01)
|
0.41 (<0.01)
|
0.16 (0.10)
|
-0.24 (0.01)
|
-0.34 (<0.01)
|
0.23 (0.02)
|
Cartilage Medial Femur Central
|
0.13 (0.17)
|
0.15 (0.12)
|
0.02 (0.87)
|
0.01 (0.90)
|
0.27 (<0.01)
|
0.36 (<0.01)
|
0.16 (0.09)
|
-0.31 (<0.01)
|
-0.40 (<0.01)
|
0.08 (0.42)
|
Cartilage Medial Femur Posterior
|
0.12 (0.22)
|
0.08 (0.44)
|
0.15 (0.13)
|
0.12 (0.21)
|
0.34 (<0.01)
|
0.42 (<0.01)
|
0.13 (0.19)
|
-0.33 (<0.01)
|
-0.35 (<0.01)
|
0.04 (0.67)
|
Cartilage Medial Tibial Anterior
|
0.15 (0.13)
|
0.11 (0.28)
|
0.09 (0.42)
|
0.08 (0.44)
|
0.32 (<0.01)
|
0.39 (<0.01)
|
0.13 (0.16)
|
-0.27 (<0.01)
|
-0.32 (<0.01)
|
0.03 (0.78)
|
Cartilage Medial Tibial Central
|
0.13 (0.17)
|
0.17 (0.08)
|
0.11 (0.25)
|
0.12 (0.20)
|
0.29 (<0.01)
|
0.37 (<0.01)
|
0.15 (0.11)
|
-0.31 (<0.01)
|
-0.41 (<0.01)
|
0.12 (0.21)
|
Cartilage Medial Tibial Posterior
|
-0.04 (0.68)
|
-0.07 (0.50)
|
-0.09 (0.38)
|
-0.06 (0.53)
|
0.27 (<0.01)
|
0.31 (<0.01)
|
0.07 (0.48)
|
-0.08 (0.38)
|
-0.19 (0.05)
|
0.05 (0.61)
|
BML Medial Femur Central
|
0.03 (0.77)
|
0.09 (0.34)
|
-0.05 (0.63)
|
0.03 (0.74)
|
0.17 (0.08)
|
0.22 (0.02)
|
0.07 (0.45)
|
-0.10 (0.28)
|
-0.21 (0.03)
|
0.10 (0.30)
|
BML Medial Femur Posterior
|
0.06 (0.53)
|
0.004 (0.96)
|
0.03 (0.79)
|
0.10 (0.28)
|
0.09 (0.35)
|
0.12 (0.22)
|
0.03 (0.75)
|
-0.05 (0.62)
|
-0.10 (0.29)
|
0.11 (0.24)
|
BML Medial Tibial Anterior
|
0.19 (0.05)
|
0.18 (0.07)
|
0.09 (0.38)
|
0.11 (0.28)
|
0.04 (0.65)
|
0.08 (0.43)
|
0.05 (0.64)
|
-0.05 (0.61)
|
-0.19 (0.05)
|
0.22 (0.02)
|
BML Medial Tibial Central
|
0.19 (0.05)
|
0.15 (0.12)
|
0.15 (0.13)
|
0.20 (0.04)
|
0.21 (0.03)
|
0.28 (<0.01)
|
0.09 (0.35)
|
-0.22 (0.02)
|
-0.34 (<0.01)
|
0.22 (0.02)
|
BML Medial Tibial Posterior
|
-0.04 (0.72)
|
0.02 (0.80)
|
-0.05 (0.61)
|
-0.06 (0.56)
|
0.11 (0.24)
|
0.13 (0.16)
|
0.04 (0.70)
|
-0.03 (0.73)
|
0.00 (0.97)
|
-0.01 (0.94)
|
Data shown are Spearman correlation coefficients, adjusted for age, sex, and BMI. P-values for the correlations are in parentheses. Significant p-values are presented in bold font. WORMS= Whole-organ Magnetic Resonance Imaging Score; WOMAC=Western Ontario and McMaster Universities Osteoarthritis Index; VAS=Visual Analog Scale; KL=Kellgren-Lawrence grade; JSW=Joint space width; JSN=joint space narrowing; BML=Bone Marrow Lesion.
Table 4. Baseline cartilage and BML scores in radiographic non-progressors compared to fast progressors in NYU (4A) and OAI (4B) SKOA.
Table 4A – NYU
|
Non-Progressors JSN≤0 mm (n=39) Mean (SD)
|
Fast Progressors JSN ≥0.5mm(n=45) Mean (SD)
|
P-Value
|
FDR
|
Cartilage Medial Overall (0-30)
|
13.88 (8.61)
|
15.42 (8.71)
|
0.42
|
0.42
|
Cartilage Lateral Overall (0-30)
|
3.56 (6.62)
|
5.24 (7.89)
|
0.298
|
0.398
|
BML Medial Overall (0-15)
|
0.59 (1.14)
|
1.78 (2.22)
|
0.003
|
0.014
|
BML Lateral Overall (0-15)
|
0.33 (0.80)
|
0.16 (0.47)
|
0.214
|
0.398
|
Table 4B – OAI
|
Non-Progressors JSN≤0 mm (n=29) Mean (SD)
|
Fast Progressors JSN ≥0.5mm (n=34) Mean (SD)
|
P-Value
|
FDR
|
Cartilage loss Medial (0-36)
|
4.21 (2.90)
|
7.09 (3.78)
|
0.006
|
0.014
|
Cartilage loss Lateral (0-36)
|
5.04 (4.09)
|
2.32 (2.80)
|
0.012
|
0.025
|
BML Medial Overall (0-36)
|
1.21 (1.82)
|
3.27 (2.31)
|
0.002
|
0.013
|
BML Lateral Overall (0-36)
|
2.23 (2.84)
|
0.98 (1.50)
|
0.072
|
0.102
|
Abbreviations: WORMS= Whole-organ Magnetic Resonance Imaging Score; BML=bone marrow lesion; JSN=joint space narrowing; FDR=false discovery rate. JSN values are expressed in mm as mean ± standard deviation (SD). All comparisons are versus non-progressors (JSN ≤0 mm). Significant p-values are represented in bold font.
Baseline MRI medial BML (mBML) and meniscus scores predict the progression of radiographic findings. BML detected by MRI has been associated with radiographic progression of knee OA (20-22, 43). We analyzed MRI findings in our two OA cohorts to determine whether: 1) MRI findings predicted the subset of fast progressors (JSN>0.5mm). We note that in the NYU cohort, MRI was available for the initial 111 enrollees previously described (12). Consistent with prior literature, fast progressors (≥0.5 mm) in the NYU cohort had significantly higher baseline medial BML scores (1.78±2.22) than non-progressors (0.59±1.14; p <0.014; Table 4A). Similarly, in the OAI cohort, medial BML scores were also higher in the fast-progressors cohort (1.21 vs. 3.27; p<0.013).
In the OAI cohort, “Cartilage Loss Medial” was also associated with progression, though this was not observed in the NYU cohort (Table 4B). In the combined cohort, association of medial BML alone with mJSN is also shown in Table 5 (AUC=0.59; 95%CI (0.51-0.67; p<0.035). The odds ratio (OR) for fast progression associated with medial BML was 2.43 (95% CI: 1.44-4.08; p<0.0001) Table 6). In the NYU cohort, the MRI meniscus score correlated (r=0.23; p=0.02) with mJSN (Table 3), though AUC (0.56; p=0.178) was modest and not significant in the dichotomized radiographic progression analysis (progressors vs. non-progressors). When meniscal findings were combined with BML and PBL inflammatory gene expression the AUC for either alone increased significantly to 0.73 (p<0.0001; Table 7).
Table 5. The area under the receiver operating characteristic curves (AUC) comparing baseline PBL inflammatory gene expression biomarkers, with and without medial BML, to predict 24-month fast radiographic knee OA progression in the combined NYU and OAI cohorts.
NYU + OAI Combined - Biomarkers
|
Biomarkers + Baseline Medial BML
|
Fast Progressors (JSN>0.5mm) vs.
Non-Progressors
(JSN <=0.0mm)
|
AUC
(95% CI)
|
Adjusted
P value
for model
|
Fast Progressors (JSN>0.5mm) vs.
Non-Progressors
(JSN<=0.0mm)
|
AUC
(95% CI)
|
P value for model
|
P value for Model performance Increase
|
Age
|
0.46 (0.38-0.54)
|
0.27
|
Age
|
0.62 (0.54-0.69)
|
0.005
|
0.002
|
Sex
|
0.52 (0.44-0.60)
|
0.31
|
Sex
|
0.60 (0.52-0.68)
|
0.006
|
0.002
|
BMI
|
0.54 (0.46-0.62)
|
0.27
|
BMI
|
0.61 (0.53-0.69)
|
0.005
|
0.041
|
ASB
|
0.50 (0.42 - 0.58)
|
0.498
|
ASB
|
0.59 (0.51 - 0.67)
|
0.014
|
0.016
|
COX-2
|
0.61 (0.53 - 0.68)
|
0.015
|
COX-2
|
0.65 (0.58 - 0.73)
|
<0.0001
|
0.074
|
IL-1β
|
0.57 (0.49 - 0.65)
|
0.081
|
IL-1β
|
0.62 (0.54 - 0.70)
|
0.004
|
0.088
|
TNFα
|
0.50 (0.42 - 0.58)
|
0.498
|
TNFα
|
0.59 (0.51 - 0.67)
|
0.020
|
0.021
|
IL1β + COX2
|
0.56 (0.48 - 0.64)
|
0.094
|
IL1β + COX2
|
0.61 (0.54 - 0.69)
|
0.005
|
0.085
|
IL1β + TNFα
|
0.52 (0.43 - 0.60)
|
0.381
|
IL1β + TNFα
|
0.60 (0.52 - 0.68)
|
0.011
|
0.024
|
COX2 + TNFα
|
0.57 (0.49 - 0.65)
|
0.081
|
COX2 + TNFα
|
0.62 (0.54 - 0.69)
|
0.005
|
0.136
|
IL-1β + COX-2 + TNFα
|
0.62 (0.54 - 0.69)
|
0.007
|
IL-1β + COX-2 + TNFα
|
0.66 (0.58 - 0.73)
|
<0.0001
|
0.114
|
Baseline BML
|
0.59 (0.51 - 0.67)
|
0.035
|
Baseline BML
|
|
|
|
Osteophytes MFC
|
0.47 (0.39 - 0.55)
|
0.309
|
Osteophytes MFC
|
0.57 (0.49 - 0.65)
|
0.035
|
0.010
|
Osteophytes MTP
|
0.48 (0.40 - 0.56)
|
0.356
|
Osteophytes MTP
|
0.58 (0.50 - 0.66)
|
0.035
|
0.011
|
Osteophytes LFC
|
0.58 (0.50 - 0.66)
|
0.072
|
Osteophytes LFC
|
0.61 (0.53 - 0.69)
|
0.005
|
0.096
|
Osteophytes LTP
|
0.47 (0.39 - 0.55)
|
0.309
|
Osteophytes LTP
|
0.58 (0.50 - 0.66)
|
0.029
|
0.007
|
Osteophytes
MFC + MTP
|
0.46 (0.38 - 0.54)
|
0.207
|
Osteophytes MFC + MTP
|
0.57 (0.48 - 0.65)
|
0.056
|
0.006
|
Osteophytes
LFC + LTP
|
0.57 (0.49 - 0.65)
|
0.081
|
Osteophytes LFC + LTP
|
0.61 (0.53 - 0.69)
|
0.006
|
0.083
|
ALL osteophytes
|
0.57 (0.49 - 0.65)
|
0.081
|
ALL osteophytes
|
0.61 (0.53 - 0.69)
|
0.005
|
0.064
|
IL-1β + COX-2 + TNFα + osteophytes MFC + MTP
|
0.59 (0.51 - 0.67)
|
0.038
|
IL-1β + COX-2 + TNFα + osteophytes MFC + MTP
|
0.63 (0.55 - 0.71)
|
0.002
|
0.114
|
IL-1β + COX-2 + TNFα + osteophytes LFC + LTP
|
0.67 (0.59 - 0.74)
|
<0.0001
|
IL-1β + COX-2 + TNFα + osteophytes LFC + LTP
|
0.68 (0.60 - 0.76)
|
<0.0001
|
0.271
|
IL-1β + COX-2 + TNFα + ALL osteophytes
|
0.67 (0.59 - 0.74)
|
<0.0001
|
IL-1β + COX-2 + TNFα + ALL osteophytes
|
0.68 (0.61 - 0.76)
|
<0.0001
|
0.223
|
ALL markers
|
0.68 (0.61 - 0.76)
|
<0.0001
|
|
|
|
|
ASB – age, sex and BMI; PBL=peripheral blood leukocyte; BML=Bone Marrow Lesion; JSN=joint space narrowing; COX-2=cyclooxygenase-2; IL-1β=interleukin-1 beta; TNFα=tumor necrosis factor alpha; 95% CI=95% confidence intervals. Medial (MTP) and lateral (LTP) osteophytes in tibial plateau and medial (MFC) and lateral (LFC) femoral condyle. All comparisons are versus non-progressors (JSN≤0 mm). Significant p-values are represented in bold font. Total number of Progressors (n=85) and Non-Progressors (n=115) in the combined cohort.
Table 6: The odds ratios (OR) comparing baseline inflammatory PBL inflammatory gene expressionbiomarkers and osteophytes with and without medial BML to predict 24-month fast radiographic knee OA progression in the combined NYU and OAI cohorts.
Biomarkers
|
|
Biomarkers + Baseline Medial BML
|
Fast Progressors (JSN>0.5mm) vs.
Non-Progressors
(JSN <=0.0mm)
|
Odds Ratio
|
95% CI
|
Adjusted P value
for model
|
Biomarkers +
Baseline
Medial BML
|
Odds Ratio
|
95% CI
|
Adjusted P value for model
|
P value for increase
|
Age
|
1
|
(1.0 - 1.0)
|
0.769
|
Age
|
2.46
|
(1.32 - 4.6)
|
0.006
|
0.001
|
Sex
|
1
|
(1.0 - 1.0)
|
0.678
|
Sex
|
2.29
|
(1.24- 4.23)
|
0.007
|
0.002
|
BMI
|
1
|
(1.0 - 1.0)
|
0.151
|
BMI
|
2.15
|
(1.17 - 3.96)
|
0.006
|
0.005
|
COX-2
|
3.56
|
(0.92 - 13.72)
|
0.019
|
COX-2
|
8.64
|
(1.36 - 54.84)
|
<0.0001
|
<0.0001
|
IL-1β
|
1.90
|
(0.56 - 6.41)
|
0.074
|
IL-1β
|
4.09
|
(0.753 - 22.230)
|
<0.0001
|
<0.0001
|
TNFα
|
1.12
|
(0.78 - 1.63)
|
0.696
|
TNFα
|
3.11
|
(1.19 - 8.14)
|
<0.0001
|
<0.0001
|
IL1β + COX2
|
1.94
|
(0.71 - 5.27)
|
0.058
|
IL1β + COX2
|
4.03
|
(0.89 - 18.23)
|
<0.0001
|
<0.0001
|
IL1β + TNFα
|
1.37
|
(0.71 - 2.65)
|
0.212
|
IL1β + TNFα
|
3.27
|
(0.94 - 11.32)
|
<0.0001
|
<0.0001
|
COX2 + TNFα
|
1.21
|
(0.79 - 1.85)
|
0.293
|
COX2 + TNFα
|
2.74
|
(1.07 - 7.04)
|
<0.0001
|
<0.0001
|
IL-1β + COX-2 + TNFα
|
6.35
|
(0.40 - 101.55)
|
0.033
|
IL-1β + COX-2 + TNFα
|
17.37
|
(0.54 - 560.39)
|
<0.0001
|
<0.0001
|
Baseline BML
|
2.43
|
(1.44 - 4.08)
|
<0.0001
|
Baseline BML
|
|
|
|
|
Osteophytes MFC
|
1.42
|
(0.96 - 2.08)
|
0.102
|
Osteophytes MFC
|
2.82
|
(1.12 - 7.09)
|
<0.0001
|
<0.0001
|
Osteophytes MTP
|
1.01
|
(0.71 - 1.45)
|
0.884
|
Osteophytes MTP
|
3.03
|
(1.20 - 7.68)
|
<0.0001
|
<0.0001
|
Osteophytes LFC
|
1.80
|
(1.19 - 2.73)
|
0.006
|
Osteophytes LFC
|
3.82
|
(1.485 - 9.820)
|
<0.0001
|
<0.0001
|
Osteophytes LTP
|
1.25
|
(0.86 - 1.81)
|
0.226
|
Osteophytes LTP
|
3.83
|
(1.452 - 10.107)
|
<0.0001
|
<0.0001
|
Osteophytes MFC + MTP
|
1.78
|
(0.76 - 4.14)
|
0.212
|
Osteophytes MFC + MTP
|
4.72
|
(1.12 - 19.87)
|
<0.0001
|
<0.0001
|
Osteophytes LFC + LTP
|
3.35
|
(1.35 - 8.30)
|
0.002
|
Osteophytes LFC + LTP
|
9.14
|
(1.98 - 42.32)
|
<0.0001
|
<0.0001
|
ALL osteophytes
|
4.20
|
(0.60 - 29.65)
|
0.007
|
ALL osteophytes
|
12.10
|
(0.83 - 177.02)
|
<0.0001
|
<0.0001
|
IL-1β + COX-2 + TNFα + osteophytes MFC + MTP
|
11.27
|
(0.27 - 467.4)
|
0.038
|
IL-1β + COX-2 + TNFα + osteophytes MFC + MTP
|
30.58
|
(0.37 - 2532.1)
|
<0.0001
|
<0.0001
|
IL-1β + COX-2 + TNFα + osteophytes LFC + LTP
|
62.06
|
(1.15 - 3365.4)
|
<0.0001
|
IL-1β + COX-2 + TNFα + osteophytes LFC + LTP
|
232.9
|
(1.85 - 29402.2)
|
<0.0001
|
<0.0001
|
IL-1β + COX-2 + TNFα + ALL osteophytes
|
69.48
|
(0.40 - 12199.8)
|
<0.0001
|
IL-1β + COX-2 + TNFα + ALL osteophytes
|
310.3
|
(0.67 - 142727.2)
|
<0.0001
|
<0.0001
|
ALL markers
|
310.26
|
(0.67 - 142727.2)
|
<0.0001
|
|
|
|
|
NaN
|
PBL=peripheral blood leukocyte; BML=Bone Marrow Lesion; JSN=joint space narrowing; COX-2=cyclooxygenase-2; IL-1β=interleukin-1 beta; TNFα=tumor necrosis factor alpha; 95% CI=95% confidence intervals. All comparisons are versus non-progressors (JSN≤0 mm). Medial (MTP) and lateral (LTP) osteophytes in tibial plateau and medial (MFC) and lateral (LFC) femoral condyle. Total number of Progressors (n=66) and Non-Progressors (n=63) in the combined cohort
Table 7. The area under the receiver operating characteristic curves (AUC) comparing baseline PBL inflammatory gene expression biomarkers, with and without medial BML, to predict 24-month fast radiographic knee OA progression in the NYU cohort (N=111).
Biomarkers
|
Biomarkers + Baseline Medial BML
|
Fast Progressors (JSN>0.5mm) vs.
Non-Progressors
(JSN <=0.0mm)
|
AUC
(95% CI)
|
Adjusted P value for model
|
Fast Progressors (JSN>0.5mm) vs.
Non-Progressors
(JSN <=0.0mm)
|
AUC
(95% CI)
|
Adjusted P value for model
|
P value for Model performance Increase
|
Age
|
0.41 (0.28-0.53)
|
0.102
|
Age
|
0.69 (0.57-0.80)
|
0.002
|
0.001
|
Sex
|
0.53 (0.41-0.66)
|
0.305
|
Sex
|
0.62 (0.50-0.74)
|
0.024
|
0.021
|
BMI
|
0.60 (0.48-0.72)
|
0.102
|
BMI
|
0.67 (0.56-0.79)
|
0.003
|
0.102
|
ASB
|
0.56 (0.44 - 0.69)
|
0.229
|
ASB
|
0.65 (0.53 - 0.77)
|
0.015
|
0.075
|
COX-2
|
0.68 (0.56 - 0.80)
|
0.005
|
COX-2
|
0.74 (0.63 - 0.85)
|
<0.0001
|
0.063
|
IL-1β
|
0.57 (0.44 - 0.69)
|
0.229
|
IL-1β
|
0.65 (0.53 - 0.77)
|
0.015
|
0.073
|
TNFα
|
0.55 (0.42 - 0.67)
|
0.255
|
TNFα
|
0.63 (0.51 - 0.75)
|
0.020
|
0.068
|
IL1β + COX2
|
0.58 (0.46 - 0.71)
|
0.209
|
IL1β + COX2
|
0.65 (0.53 - 0.77)
|
0.015
|
0.109
|
IL1β + TNFα
|
0.56 (0.43 - 0.68)
|
0.229
|
IL1β + TNFα
|
0.64 (0.52 - 0.76)
|
0.016
|
0.073
|
COX2 + TNFα
|
0.54 (0.41 - 0.66)
|
0.287
|
COX2 + TNFα
|
0.63 (0.51 - 0.75)
|
0.020
|
0.066
|
IL-1β + COX-2 + TNFα
|
0.67 (0.55 - 0.79)
|
0.009
|
IL-1β + COX-2 + TNFα
|
0.74 (0.63 - 0.85)
|
<0.0001
|
0.038
|
Baseline BML
|
0.65 (0.53 - 0.77)
|
0.020
|
Baseline BML
|
0.65 (0.53 - 0.77)
|
0.015
|
1.000
|
Osteophytes MFC
|
0.56 (0.44 - 0.68)
|
0.229
|
Osteophytes MFC
|
0.63 (0.51 - 0.75)
|
0.020
|
0.075
|
Osteophytes MTP
|
0.58 (0.45 - 0.70)
|
0.220
|
Osteophytes MTP
|
0.63 (0.51 - 0.76)
|
0.020
|
0.128
|
Osteophytes LFC
|
0.57 (0.44 - 0.69)
|
0.229
|
Osteophytes LFC
|
0.64 (0.52 - 0.76)
|
0.017
|
0.095
|
Osteophytes LTP
|
0.53 (0.41 - 0.66)
|
0.309
|
Osteophytes LTP
|
0.65 (0.53 - 0.77)
|
0.015
|
0.034
|
Osteophytes MFC + MTP
|
0.55 (0.42 - 0.68)
|
0.249
|
Osteophytes MFC + MTP
|
0.61 (0.49 - 0.74)
|
0.035
|
0.095
|
Osteophytes LFC + LTP
|
0.59 (0.46 - 0.71)
|
0.209
|
Osteophytes LFC + LTP
|
0.65 (0.53 - 0.77)
|
0.015
|
0.072
|
ALL osteophytes
|
0.57 (0.44 - 0.69)
|
0.229
|
ALL osteophytes
|
0.62 (0.49 - 0.74)
|
0.035
|
0.101
|
IL-1β + COX-2 + TNFα + osteophytes MFC + MTP
|
0.68 (0.57 - 0.80)
|
0.005
|
IL-1β + COX-2 + TNFα + osteophytes MFC + MTP
|
0.72 (0.61 - 0.83)
|
<0.0001
|
0.126
|
IL-1β + COX-2 + TNFα + osteophytes LFC + LTP
|
0.70 (0.59 - 0.82)
|
0.003
|
IL-1β + COX-2 + TNFα + osteophytes LFC + LTP
|
0.75 (0.64 - 0.85)
|
<0.0001
|
0.074
|
IL-1β + COX-2 + TNFα + ALL osteophytes
|
0.70 (0.58 - 0.81)
|
0.003
|
IL-1β + COX-2 + TNFα + ALL osteophytes
|
0.73 (0.62 - 0.84)
|
<0.0001
|
0.135
|
IL-1β + COX-2 + TNFα + BML + ALL osteophytes
|
0.73 (0.62 - 0.83)
|
<0.0001
|
IL-1β + COX-2 + TNFα + BML + ALL osteophytes
|
0.73 (0.62 - 0.84)
|
<0.0001
|
1.000
|
Meniscus sum
|
0.55 (0.43 - 0.68)
|
0.244
|
Meniscus sum
|
0.63 (0.51 - 0.75)
|
0.020
|
0.084
|
IL-1β + COX-2 + TNFα + Meniscus sum
|
0.69 (0.58 - 0.81)
|
0.003
|
IL-1β + COX-2 + TNFα + Meniscus sum
|
0.73 (0.62 - 0.83)
|
<0.0001
|
0.152
|
ASB- age, sex and BMI; PBL=peripheral blood leukocyte; BML=Bone Marrow Lesion; JSN=joint space narrowing; COX-2=cyclooxygenase-2; IL-1β=interleukin-1 beta; TNFα=tumor necrosis factor alpha; 95% CI=95% confidence intervals. Medial (MTP) and lateral (LTP) osteophytes in tibial plateau and medial (MFC) and lateral (LFC) femoral condyle. All comparisons are versus non-progressors (JSN≤0 mm). Total number of Progressors (n=44) and Non-Progressors (n=39).
Combinatorial biomarkers (medial BML scores and PBL gene expression) enhance prediction of radiographic progression in the combined NYU and OAI cohorts. Having shown that an inflammatory PBL transcriptome and BML by MRI individually predicted radiographic progression in our cohorts, we next determined whether any combination of biomarkers had greater predictive value than a single biomarker alone in the combined cohort.
We first examined the predictive value of the combination of baseline medial BML scores and PBL inflammatory gene expression markers by AUC analyses. As shown in Table 5, the association of age, sex and BMI alone with fast progressors (JSN ≥0.5 mm) was not significant with AUC ranging from 0.46 to 0.54. Moreover, causal analysis (Figure 2) did not show a direct effect of these variables on joint space narrowing. The association of BML alone with fast progressors (JSN ≥0.5 mm) in the combined cohort was AUC=0.59; (95%CI (0.51-0.67), p =0.035; Table 5). In each instance, the combination of medial BML and individual PBL inflammatory gene expression increased the predictive value of either biomarker alone. Specifically, the combination of medial BML and PBL COX-2 expression (AUC 0.65, p <0.0001) yielded the maximal predictive power for fast progressors (Table 5). Additionally, the increased AUC model performance was significant for PBL mRNA transcripts (COX-2, IL1β, and TNFα) in combination with BML (Table 5) in predicting radiographic progression.
In addition to calculating the AUCs, we also determined the Odds Ratio (OR) for progression for each biomarker, alone and in combination. As shown in Table 6, the OR for baseline PBL COX-2 and IL-1β predicting fast progressors (JSN≥ 0.5 mm) were 3.56 and 1.90 respectively, and a combination of all three inflammatory PBL gene expression biomarkers (IL-1 β + TNFα + COX-2) had an OR of 6.35. The addition of baseline medial BML to inflammatory PBL IL-1β, COX-2, and TNFα markers increased the OR significantly to 17.37 (p <0.0001).
Baseline radiographic osteophytes scores and progression of joint space narrowing. Since osteophytes have also been reported to associate with radiographic progression in knee OA (44-47), we added osteophyte assessment to our predictive models in the combined cohort. As shown in Table 6, combination of medial and lateral osteophyte scores increased the OR for progression to 4.2 (p=0.007). The combination of the PBL inflammatory gene expression and osteophytes markers increased the OR to 69.48. (p<0.0001), while the combination of baseline BML, osteophytes, and molecular biomarkers further increased OR to 310.30 (p<0.0001). Similarly, AUC analysis indicated that the predictive value of osteophytes alone significantly increased when combined with PBL gene expression or BML (Table 5).
Causal graph analysis. To further explore interactions of baseline radiographic (osteophytes) MRI features (medial BML, cartilage, and meniscus), and PBL inflammatory gene expression biomarkers, we performed causal graph analysis to determine the inter-dependent pathways of these factors on continuous JSN over 24 months (Figure 2). Baseline PBL COX-2 gene expression and baseline medial BML each independently played a causal role or positively influenced JSN. Moreover, age, sex and BMI did not show a direct effect on joint space narrowing. However, the open circles adjacent to several predictors, including sex, age, BMI, and osteophytes, indicate that there may be hidden confounders, which may influence their relationships with JSN. These new data demonstrate that the PBL inflammatory COX-2 gene expression and medial BML, independently from other MRI or radiographic features, influence radiographic JSN progression.