PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Osteoarthritis Cartilage. Author manuscript; available in PMC 2013 November 1.
Published in final edited form as:
PMCID: PMC3608522
NIHMSID: NIHMS404905

Association between serum and urine biomarkers and lumbar spine individual radiographic features: The Johnston County Osteoarthritis Project

Adam P. Goode, PT, DPT, PhD, Assistant Professor, Stephen W. Marshall, PhD, Professor, Virginia B. Kraus, MD, PhD, Professor, Jordan B. Renner, MD, Professor, Til Stürmer, MD, MPH, PhD, Professor, Timothy S. Carey, MD, MPH, Professor, Debra E. Irwin, PhD, MSPH, Research Associate Professor, and Joanne M. Jordan, MD, MPH, Professor

Abstract

Objective

1) To determine associations between radiographic features of lumbosacral spine disc space narrowing (DSN) and osteophytes (OST) and joint metabolism biomarkers (serum cartilage oligomeric matrix protein (COMP), hyaluronic Acid (HA), collagen neoepitope (C2C), C-propeptide of type II procollagen (CP-II), urine C-terminal cross-linking telopeptide (CTX-II) and N-terminal telopeptide (NTX-I)). 2) To explore interactions with race, gender and low back symptoms.

Design

Cross-sectional analysis of 547 participants enrolled in the Johnston County Osteoarthritis Project from 2003–2004. Mean biomarker levels were estimated with linear regression. Proportional and partial proportional odds models were used to estimate associations. Interactions were tested with likelihood ratio tests at a p-value<0.10. Biomarkers were natural log (ln) transformed.

Results

Significant differences in mean biomarker levels were found across severity of DSN for lnHA and lnC2C and lnCTX-II across severity of both DSN and OST. Moderate to strong associations were found between biomarkers of type II collagen and DSN, whereas associations with OST were weak. An association between lnHA and DSN was seen in women (adjusted odds ratio [aOR]=1.34 (95% CI 1.08, 1.65)) but no association among men (aOR=0.90 (95% CI 0.63, 1.26)). In Caucasians there was a decreased association with NTX-I and OST (aOR=0.67 (95% CI 0.49, 0.91)) and no association in African Americans (aOR=1.06 (95% CI 0.76, 1.47)). There was a positive association of lnCOMP with DSN among those with low back symptoms (aOR=1.82 (95% CI 1.02, 3.27)), but no association in those without low back symptoms (aOR=0.65 (95% CI 0.35, 1.20)).

Conclusion

Joint metabolism biomarkers suggest biological differences in the pathologic process involved in DSN and OST that may be gender (HA) and ethnicity (NTX-I) specific.

Introduction

Intervertebral disc degeneration is a common condition with prevalence estimates from community-based studies to be between 50–64% (mean age >65 years).13 The degenerative process is characterized by loss of water, aggrecan and types I and II collagen,4,5 resulting in disc space narrowing (DSN) on plain film radiographs. Vertebral osteophytes (OST) are likely a response to result from cartilage loss, biomechanical stress or trauma to the annular insertion.6 Vertebral OST prevalence estimates from community-based studies (mean age >65 years) range between 75–94%.13

Joint metabolism biomarkers, found in serum and urine, are described according to the tissue(s) in which the markers are abundant and with respect to the molecules they were developed to identify. Type I collagen markers, such as the N-terminal telopeptide (NTX-I), reflect bone turnover or resorption. Type II collagen markers, the C-terminal cross-linked telopeptide (CTX-II) and the collagenase-generated neoepitope (C2C) are biomarkers of type II collagen degradation, whereas C-propeptide of type II procollagen (CPII) reflects cartilage synthesis.7 The non-aggrecan and non-collagenous marker, such as cartilage oligomeric matrix protein (COMP), reflects cartilage degradation but may also be found in bone, ligaments, tendons and vascular smooth muscle.8 Hyaluronic Acid (HA) is a glycosaminoglycan that is indicative of osteoarthritis (OA) and synovial inflammation. 7,9 These particular biomarkers have been studied in OA of the hip1012 and knee1319. Only CTX-II has been evaluated for an association with radiographic changes in the lumbar spine.20,21

Our previous work with the Johnston County (JoCo) OA Project has identified that the lumbar spine individual radiographic features (IRFs) of DSN and OST differ substantially from one another in prevalence suggesting a different pathophysiologic process between the two. Furthermore, we found that the associations with demographic, clinical and radiographic knee, hip and hand OA differ between these lumbar spine IRFs.22 Our primary objective for this analysis was to compare a broad spectrum of serum (COMP, HA, C2C, CP-II) and urine (NTX-I, CTX-II) biomarkers to determine if they would reflect one or both of the osteoarthritis-related processes represented by lumbar spine DSN and OST. Our second objective was to determine if the associations of biomarkers with lumbar spine radiologic features would differ among race and gender groups and according to the presence of self-reported low back symptoms. We hypothesized that these biomarkers would be independently associated with lumbar spine DSN and OST above and beyond known associations with concomitant appendicular joint radiographic OA.

Methods and Materials

Participants

Data for these cross-sectional analyses come from the JoCo OA Project, an ongoing population-based study set in 6 rural townships of Johnston County, North Carolina. The purpose of the JoCo OA Project is to determine the incidence, prevalence and progression of knee, hip, hand, and spine OA. The sampling strategy and recruitment methods used for the JoCo OA Project are described in detail elsewhere.23,24 Briefly, baseline enrollment (1991–98) included civilian, non-institutionalized, African American or Caucasian adult participants ages 45 years and older recruited by probability sampling. The participants for these particular analyses come from 1,015 participants enrolled in the JoCo OA Project during cohort enrichment (T1*, 2003–04), following the first follow-up (T1, 1999–03) of participants initially recruited from 1991–98. The T1* cohort enrichment aimed to increase the sample for African Americans (AAs) and younger participants who were lost over the follow-up period from the study initiation. As such, participants at T1* were younger (mean age 59.3 vs. 65.8 years) and had a higher proportion of AAs (40% vs. 28%) than those at T1; the 2 groups did not differ according to gender.23

Radiographic evaluation

Of the 1,015 participants that underwent clinical examination, 840 had complete radiographic data for DSN and OST. Reasons for not having lumbar spine radiographs included participant refusals (n=6), women of reproductive age (<50 years of age) excluded by protocol from having lumbar spine radiographs (n=132), participants exceeding radiographic table weight limit (n=23) or missing (n=6). Lateral lumbar spine films were taken with the participant lying on his/her left side. All lateral lumbar spine radiographs were graded at each lumbar level by a single bone and joint radiologist (JBR) without regard to participants’ biomarker levels. DSN and OST were graded based upon the Burnett Atlas25 in a semi-quantitative fashion (0=none, 1=mild, 2=moderate and 3=severe). The grading for OST was done for each superior and inferior aspect of the anterior face of the lumbar vertebra.

Participants completed weight bearing posterior-anterior knee radiography of both knees (n=979) with a Synaflexer (CCBR-Synarc, San Francisco, CA) positioning device, and bilateral hip radiography (n=830) with supine anterior-posterior pelvis radiographs. The presence of a knee prosthesis due to arthroplasty was the primary reason for missing knee radiographic data. Women of reproductive age (<50 years, n=132) were not subjected to pelvic radiation; this accounted for the majority of missing hip radiographic data. Posterior-anterior hand radiographs were available for 1,012 participants and were obtained with the beam focused on the 3rd metacarpal-phalangeal joint; a total of 30 hand joints bilaterally were graded including: the distal interphanlangeal [DIP], proximal interphalangeal [PIP], metacarpal phalangeal [MCP], carpometacarpal [CMC] and thumb IP and MCP joints. All hip, knee and hand radiographs were read for Kellgren-Lawrence (K-L)26 score by a single bone and joint radiologist (JBR). Inter-rater and intra-rater reliability have been reported previously with a weighted kappa of 0.86 and 0.89 for both the hip and knee.27 Hip and knee OA, for these analyses, was defined as a K-L score of 2–4 in at least one extremity. Hand OA was defined, similar to a previous definition, as having at least one hand with a K-L grade of 2–4 in one distal interphalangeal joint and in at least 2 other interphalangeal joints or carpometacarpal joints across both hands; participants were excluded on the basis of any MCP with swelling of grade 2 or higher on clinical examination.28

Biomarkers

The participants for this biomarkers study came from those participants in T1* who also had serum and urine analyzed for joint metabolism biomarkers as part of the sub-study of the JoCo OA entitled the JoCo OA Metals Exposure Sub-study. This sub-study was designed to consist of 1700 consecutive individuals from T1* and from the original cohort’s first follow-up (T1) who returned for their second follow-up (T2) in 2006–2008.29 The sample for the current analysis included participants selected from T1* who had complete biomarker analyses performed (n=555). Urinary CTX-II had missing values for 18 participants and urinary NTX-I had a single participant missing value. There were no missing serum biomarker values; however, some participants with biomarker values had missing or unreadable (congenital defect or surgery) lumbar spine radiographs. Therefore, the final sample for these analyses consists of 547 participants that had both complete lumbar spine radiographs and biomarker values from the original T1* enrollment (Figure).

Figure
Flow diagram of participants with complete radiographic and biomarker data for this analysis from the Johnston County Osteoarthritis (JoCo) Project new enrollment 2003 – 2004.

After collection at the clinic visit, blood and sera were separated and stored on ice. Sera were frozen within 8 hours of collection at −20° Celsius, then transferred for long-term storage to an −86° Celsius environment.

Demographic data

At the time of interview, the following data were collected: age; body mass index (BMI) calculated from height measured without shoes and weight measured with a balance beam scale, race (Caucasian/AA); gender; and low back symptoms by asking participants to answer “yes” or “no” to “On most days do you have pain, aching or stiffness in your lower back?”

Definition of biomarkers

Biomarker analyses were conducted at the laboratory of Dr. Virginia Kraus at Duke University Medical Center. Serum HA was measured with the Hylauronic Acid Test kit (Corgenix, Westmister, CO). The precision is 3.6–4.7% intra-assay variability and between 5.7–7.0% inter-assay variability. Serum COMP was measured using an in-house sandwich enzyme linked immunosorbent assay (ELISA). The reported precision is 5.8–6.6% intra-assay variability and between 8.7–9.7% inter-assay variability [REF]. C2C was measured with a competitive ELISA (Ibex, Montreal, CA). The minimum detectable concentration of C2C is 7.3 ng/ml. Intra- and inter-assay precision are 2.41% and 9.49%, respectfully. CPII was also measured with a competitive ELISA (Ibex, Montreal, CA). The minimum detectable concentration is reported to be 35.1 ng/ml and intra- and inter assay precision are 3.68% and 9.08%, respectfully. Urinary bone degradation markers, CTX-II (IDS, Boldon, UK) and NTX-I (Osteomark, Princeton, NJ) were measured with a competitive ELISA. The minimum detectable concentration of urinary CTX-II is reported to be 0.20 ng/ml and the intra- and inter-assay precision are 5.86% and 9.97%, respectfully. Urinary CTX-II and NTX-I values were corrected for creatinine concentrations measured by (Quidel, San Diego, CA).

Definition of lumbar spine radiographic features

Since no universal method exists for coding IRF in the lumbar spine, we utilized a summary procedure similar to a previous study.9 For DSN, the scoring was based upon a 0 to 15 scale by summing the values of grades 0–3 for each individual lumbar level. For example, a participant would receive a score of 0 if there were no DSN at a single level, a score of 1 if there was mild DSN at a single level, a score of 2 if there was moderate DSN at any one level or any two levels of mild DSN and so forth; a maximum score of 15 is possible representing a grade of 3 (severe DSN) at all 5 lumbar levels. These scores were collapsed into 4 mutually exclusive categories based upon the empirical distribution of severity grades in order to reflect greater lumbar spine degenerative severity with increasing scores (0=scores of 0, 1=scores of 1 and 2 combined, 2=scores of 3 and 4 combined and 3=scores of 5+ combined).

A similar scoring system was used for superior and inferior (anterior) OST with values that could range from 0 to 30. These scores were also collapsed into 4 mutually exclusive categories (0=scores of 0–2, 1=scores from 3 to 7 combined, 2=scores from 8–12 combined and 3=scores of 13+).

Statistical Analysis

Each biomarker demonstrated a right skewed distribution, so biomarkers were naturally log (ln) transformed. Descriptive statistics were generated for each outcome in the form of means and standard deviations for continuous covariates and counts and proportions for categorical covariates. Adjusted mean biomarker levels were estimated using linear regression. Analysis of variance was used to determine differences across categories of severity and adjusted mean levels of each biomarker.

Due to ordinal nature of the outcome variables, proportional odds (polytomous ordinal) logistic regression models were used for all analyses, in which the outcomes met model fit assumptions, defined as having a score test30 of p>0.05 and with graphical assessment of trends in the log-odds with incremental coding of the outcomes 31,32. The proportional odds model assumes that the odds ratios are the same for all possible cut-points dichotomizing the ordinal dependent variable (radiographic OA, for categories see Table 1) into high versus low. For violations, an unconstrained partial-proportional odds model has been recommended since it relaxes the strict proportional odds assumption for only those covariates not meeting the assumption.33 This model, therefore, provides estimates for k-1 levels of the outcome, where k is equal to the number of categories in the outcome, to best represent the non-proportional effect across levels of the outcome when a single estimate may not.31 An unconstrained partial-proportional odds model is considered a “nested” version of the proportional odds logistic regression model31; therefore, sensitivity analysis was conducted between models where this approach was required. Proportional odds ratios and their 95% confidence intervals (CI) were estimated from both models.

Table 1
Demographics and characteristics of the available 547 participants with lumbar spine radiographic data*.

All models were adjusted for age, race, gender and BMI, as these covariates have been associated with lumbar spine IRF and at least some of our selected biomarkers in previous literature.20,21 The confounding effect of the presence of knee, hip or hand OA was explored using a backward stepwise deletion with a change in estimate approach (difference in the model estimate once the covariate was removed).34 Therefore, knee, hip or hand OA were retained in the final models if there was a change in the estimate of the logit of >5% between each of the biomarkers and either DSN or OST. To check for independence of associations, each model included the other radiographic feature. Pair-wise interaction terms between each covariate and biomarker were used to assess effect measure modification. Interaction terms were retained in the model if they had a likelihood ratio test p-value of <0.10. All variables were examined for collinearity using variance inflation factors and tolerance values. No outliers were identified in the final models with sensitivity analyses utilizing leverage plots and removal of potential outliers. All analyses were conducted in Stata 10 (Stata Corp, College Station, TX).

Results

Basic demographics, clinical characteristics, and presence of knee, hip, or hand OA for each lumbar spine score category of DSN and OST are summarized in Table 1. Compared with men, women were slightly more likely to have severe DSN. In contrast, men were more likely to have severe OST. The prevalence of low back symptoms increased monotonically across severity categories of DSN but remained relatively constant across levels of OST severity. The frequency of knee, hip and hand OA increased with increasing severity of both DSN and OST. The mean age of participants was higher in categories of more severe DSN and OST. Mean BMI increased as severity of OST increased whereas remained relatively constant across DSN severity.

Table 2 summarizes the adjusted geometric means and their 95% confidence intervals of biomarkers and DSN and OST. All models warranted adjustment for knee OA whereas hip and hand OA did not result in significant confounding. The exception was for NTX-I, which required no adjustment for any appendicular joint radiographic OA. Differences in mean levels across severity of DSN were found for lnHA and lnC2C (both adjusted for knee OA) but no differences were observed across categories of OST. lnCTX-II demonstrated differences in mean levels across both DSN and OST. No significant differences were found for adjusted mean levels of lnCPII, lnCOMP or lnNTX-I for DSN or OST.

Table 2
Adjusted geometric mean levels and 95% confidence intervals by disc space narrowing and osteophytes for each biomarker.

Table 3 describes the adjusted associations between biomarkers and both DSN and OST. Identical to the adjusted mean levels, knee OA warranted inclusion as an adjustment variable in all models except with NTX-I. The strongest adjusted associations were between DSN and lnC2C, lnCPII and lnCTX-II. The relation between lnHA and DSN did not meet model fit assumptions (p=0.001), indicating a violation of the proportional odds assumption, with monotonically increasing odds ratios from 1.10 (95% CI 0.91, 1.33), 1.38 (95% CI 1.08, 1.78) and 2.18 (95% CI 1.45, 3.29) over categories of DSN. The adjusted associations between biomarkers and OST were weaker when compared to DSN. The estimate for lnNTX-I was on the opposite side of the null for OST when compared to DSN. lnCTX-II demonstrated the strongest association with OST and had a similar strength of association as DSN. However, lnCTX-II did not meet proportional odds model fit assumptions (p=0.001) and the adjusted association was much stronger in the most severe category of OST (OR=2.82 (95% CI 1.61, 4.94) compared to categories of lesser severity (OR=1.42 (95% CI 1.02, 1.99) and (OR=1.47 (95% CI 1.10, 1.97), respectively. lnCOMP was not significantly associated with either DSN or OST.

Table 3
Adjusted odds ratios with 95% confidence intervals, p-values and sample sizes for each biomarker by lumbar disc space narrowing and anterior vertebral osteophytes.

Most biomarkers did not demonstrate significant differences across race, gender or low back symptoms. The association between lnNTX-I and OST was different across race; we observed a significant inverse association of lnNTX-1 and OST (OR=0.67; 95% CI 0.49, 0.91) in Caucasians but not in AAs (OR=1.06; 95% CI 0.76, 1.47). The association between lnNTX-I and OST was also different by gender with a significant inverse association among women (OR=0.65; 95% CI 0.47, 0.89) but not among men (OR=1.12; 95% CI 0.79, 1.60). The association between lnHA and DSN had a significant moderate association with women (OR=1.34; 95% CI 1.08, 1.65) while men had no association (OR=0.90; 95% CI 0.63, 1.26). Among those with low back symptoms (n=265) there was a strong positive association with lnCOMP and DSN (OR=1.82; 95% CI 1.02, 3.27) that was not observed in participants without low back symptoms (OR=0.65; 95% CI 0.35, 1.20).

Discussion

Our study examined the relationship between a broad spectrum of biomarkers and lumbar spine IRF within a large bi-racial community-based cohort. Adding face validity to the results, in several cases, the biomarkers reflected spine pathology in a biologically plausible manner based on their known relative abundance in spinal tissues. For instance, markers of type II collagen, the most abundant protein in the nucleus pulposus20, demonstrated moderate-to-strong associations with DSN above and beyond known associations with radiographic knee OA. In contrast, the association between type II collagen biomarkers and OST were weaker. This may be due to the expected lack of type II collagen in a mature (bony) osteophyte since when the osteophyte is first laid down it is a chondrophyte with collagen II, then it undergoes endochonral ossification involving chondral anlage resorption and remodeling into a bony osteophyte that would be visible and gradable on a radiograph. In contrast, markers of less abundant non-collagenous extracellular matrix components (lnHA and lnCOMP) had weak associations with DSN but were responsive to differences in gender and low back symptoms. Furthermore, lnHA demonstrated increased adjusted mean levels with increasing severity of DSN.

Two previous cross-sectional studies have examined the role of CTX-II as a burden of disease biomarker in the spine.20,21 Our study findings are generally consistent with both of these studies regarding increasing levels of CTX-II with severity of DSN. However, our results differ from Garnero and colleagues20 that found no association between CTX-II and vertebral OST. Differences in sample demographics may be one explanation for this difference. We found that the most severe category of OST had a strong association with CTX-II. This category of OST had a higher proportion of men, who had more severe OST. This may help explain the difference between our studies, since their study consisted of women only. Other reasons may include coding scheme or grading scale differences for the lumbar spine radiographic features between studies.

Our findings indicate that associations between some, but not all, lumbar spine IRF and biomarkers differ by subgroups of race and gender. Previous studies have found that women are more likely to have DSN compared to men.1,2 In this study, a moderate independent association existed between lnHA and DSN in women. Therefore, lnHA reflects this previously known gender difference in DSN. In addition, our results indicate that there are significant differences in association across subgroups of both gender and race between lnNTX-I and OST. In race and gender subgroups, lnNTX-I reflected a decreased odds of OST. Women are known to have less severe vertebral OST when compared to men22 and this could be one explanation for the findings here with lnNTX-I. The differences in association we found between lnNTX-I and OST across race may also be explained by bone mineral content since racial differences in this process have been established.37 Although the race and gender differences observed in this study are consistent with differences in bone and spine structure, we were unable to control for the established relationship between decreased bone mineral density and lnNTX-I, which may confound our reported associations.

Previous studies have found a relationship between biomarkers and both hip and knee symptoms38,39, and low back and sciatic symptoms40. We explored the relationship between biomarkers and low back symptoms. Previous work by us and others has revealed a modest association between low back symptoms and DSN.13,22 Furthermore, the association between DSN and low back symptoms has been found to increase with severity and number of affected lumbar vertebral levels.1 Due to the multidimensional and complex nature of low back pain, there are challenges to interpreting differences in associations of biomarkers with low back symptoms. However, the association is noteworthy between lnCOMP and DSN among those with and without low back symptoms. Only among participants with low back symptoms, was there a strong direct association between lnCOMP and DSN. Therefore, it is possible that lnCOMP reflects the ongoing degeneration of the intervertebral disc reflected by DSN and resulting in associated symptoms that have been reported with this process.

The confounding effect of concomitant peripheral joint OA may differ based upon specific joint site and the biomarker analyzed.9 Interestingly, lnNTX-I, associated with DSN, demonstrated no change in estimate when controlling for knee, hip or hand OA. This suggests that these effects may work through a different pathologic process. Hassett and colleagues35 reported strong associations between both knee and hip OA progression and DSN while no association was found with hand OA. In agreement, none of our analyses were confounded by hand OA, but several required adjustment for knee OA. The association between lnCOMP, lnCTX-II, lnCPII, lnC2C and lnHA with knee OA has been established in previous literature36; therefore, it is not surprising that models with these biomarkers required adjustment due to confounding by the presence of knee OA. Hip OA was not a confounder in any models, and this is consistent with our previous work that showed that hip OA is not independently associated with lumbar spine IRF.22 This further strengthens the argument that hip OA follows a different etiologic process from that of the lumbar spine. These findings are similar to the previous work by Kraus and colleagues9, as they also found differences in the confounding effect of concomitant peripheral joint OA and subsequent adjustment in multivariate models.

A major strength of these analyses is that they were conducted in a well-defined, community-based population in which we examined a broad spectrum of biomarkers across a balanced representation of AAs and Caucasians and both men and women. Moreover, we were able to analyze LS spine associations with biomarkers in the context of concomitant knee, hip, and hand radiographic OA. The primary limitation of this study was its cross-sectional design; thus we could not address the temporal relationship between the onset of serum biomarker abnormalities and onset of OA; this can only be assessed using longitudinal analyses. Our study population does not reflect the general population in so far as our cohort contained a greater proportion of African Americans and had a younger mean age. We adjusted for hip, knee, and hand OA to take into account their potential contribution to biomarker levels; however, we did not control for such things as medication use, liver function, and kidney function, which may affect serum levels of biomarkers. All our variables had attendant measurement error. Assuming independent, non-differential errors, our estimates of associations would be an underestimation of the true associations. Despite these limitations, we conclude that our findings indicate that these bone and joint biomarkers reflect differences in the process of degeneration between DSN and OST. Furthermore, some of these biomarkers may show differences in lumbar spine degeneration in both race and gender subgroups and the presence of low back symptoms. The ability of biomarkers to independently reflect metabolic processes in the lumbar spine provides the impetus for future studies to determine if biomarkers can also predict degenerative changes in the lumbar spine.

Acknowledgments

The primary author would like to acknowledge the Foundation for Physical Therapy for tuition support during his PhD training at the UNC Gillings School of Global Public Health.

The Johnston County Osteoarthritis Project is supported in part by cooperative agreements S043, S1734, and S3486 from the Centers for Disease Control (CDC) and Prevention/Association of Schools of Public Health; the NIAMS Multipurpose Arthritis and Musculoskeletal Disease Center grant 5-P60-AR30701; the NIAMS Multidisciplinary Clinical Research Center grant 5 P60 AR49465-03. Salary support for VBK is provided by NIH/NIA OAIC 5P30 AG028716 and NIH/NIAMS 5P01-AR050245. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC, NIA or NIAMS.

Dr. Til Stürmer receives investigator-initiated research funding and support as Principal Investigator (RO1 AG023178) and Co-Investigator (RO1 AG018833) from the National Institute on Aging at the National Institutes of Health. He also receives research funding as Principal Investigator of the UNC-DEcIDE center from the Agency for Healthcare Research and Quality. Dr. Stürmer does not accept personal compensation of any kind from any pharmaceutical company, though he receives salary support from the UNC Center of Excellence in Pharmacoepidemiology and Public Health and from unrestricted research grants from pharmaceutical companies to UNC.

Footnotes

Contributions:

All authors of this work have made substantial contributions to the conception and design, acquisition of data, analysis, interpretation, drafting of manuscript and final approval of submission.

Competing Interest statement:

All authors disclose they have no financial or personal relationships with other people or organizations that could potentially and inappropriately bias their work and conclusions.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Contributor Information

Adam P. Goode, Department of Community and Family Medicine, Duke University Medical Center.

Stephen W. Marshall, Department of Epidemiology, Gillings School of Global Public Health, University of North Carolina.

Virginia B. Kraus, Department of Rheumatology, Duke University Medical Center.

Jordan B. Renner, Department of Radiology, University of North Carolina.

Til Stürmer, Department of Epidemiology, University of North Carolina.

Timothy S. Carey, Director Cecil G. Sheps Center, University of North Carolina.

Debra E. Irwin, Department of Epidemiology, University of North Carolina.

Joanne M. Jordan, Department of Medicine, Orthopedics, and Epidemiology, Division of Rheumatology, Allergy and Immunology, University of North Carolina.

References

1. de Schepper EI, Damen J, van Meurs JB, et al. The association between lumbar disc degeneration and low back pain: the influence of age, gender, and individual radiographic features. Spine (Phila Pa 1976) 2010;35(5):531–536. [PubMed]
2. Pye SR, Reid DM, Smith R, et al. Radiographic features of lumbar disc degeneration and self-reported back pain. J Rheumatol. 2004;31(4):753–758. [PubMed]
3. Muraki S, Oka H, Akune T, et al. Prevalence of radiographic lumbar spondylosis and its association with low back pain in elderly subjects of population-based cohorts: the ROAD study. Ann Rheum Dis. 2009;68(9):1401–1406. [PubMed]
4. Antoniou J, Steffen T, Nelson F, et al. The human lumbar intervertebral disc: evidence for changes in the biosynthesis and denaturation of the extracellular matrix with growth, maturation, ageing, and degeneration. J Clin Invest. 1996;98(4):996–1003. [PMC free article] [PubMed]
5. Eyre DR, Muir H. Quantitative analysis of types I and II collagens in human intervertebral discs at various ages. Biochim Biophys Acta. 1977;492(1):29–42. [PubMed]
6. Pye SR, Reid DM, Lunt M, Adams JE, Silman AJ, O’Neill TW. Lumbar disc degeneration: association between osteophytes, end-plate sclerosis and disc space narrowing. Ann Rheum Dis. 2007;66(3):330–333. [PMC free article] [PubMed]
7. Bauer DC, Hunter DJ, Abramson SB, et al. Classification of osteoarthritis biomarkers: a proposed approach. Osteoarthritis Cartilage. 2006;14(8):723–727. [PubMed]
8. Muller G, Michel A, Altenburg E. COMP (cartilage oligomeric matrix protein) is synthesized in ligament, tendon, meniscus, and articular cartilage. Connect Tissue Res. 1998;39(4):233–244. [PubMed]
9. Kraus VB, Kepler TB, Stabler T, Renner J, Jordan J. First qualification study of serum biomarkers as indicators of total body burden of osteoarthritis. PLoS One. 2010;5(3):e9739. [PMC free article] [PubMed]
10. Kelman A, Lui L, Yao W, Krumme A, Nevitt M, Lane NE. Association of higher levels of serum cartilage oligomeric matrix protein and N-telopeptide crosslinks with the development of radiographic hip osteoarthritis in elderly women. Arthritis Rheum. 2006;54(1):236–243. [PubMed]
11. Garnero P, Conrozier T, Christgau S, Mathieu P, Delmas PD, Vignon E. Urinary type II collagen C-telopeptide levels are increased in patients with rapidly destructive hip osteoarthritis. Ann Rheum Dis. 2003;62(10):939–943. [PMC free article] [PubMed]
12. Clark AG, Jordan JM, Vilim V, et al. Serum cartilage oligomeric matrix protein reflects osteoarthritis presence and severity: the Johnston County Osteoarthritis Project. Arthritis and rheumatism. 1999;42(11):2356–2364. [PubMed]
13. Bettica P, Cline G, Hart DJ, Meyer J, Spector TD. Evidence for increased bone resorption in patients with progressive knee osteoarthritis: longitudinal results from the Chingford study. Arthritis Rheum. 2002;46(12):3178–3184. [PubMed]
14. Garnero P, Ayral X, Rousseau JC, et al. Uncoupling of type II collagen synthesis and degradation predicts progression of joint damage in patients with knee osteoarthritis. Arthritis Rheum. 2002;46(10):2613–2624. [PubMed]
15. Lohmander LS, Yoshihara Y, Roos H, Kobayashi T, Yamada H, Shinmei M. Procollagen II C-propeptide in joint fluid: changes in concentration with age, time after knee injury, and osteoarthritis. J Rheumatol. 1996;23(10):1765–1769. [PubMed]
16. Sugiyama S, Itokazu M, Suzuki Y, Shimizu K. Procollagen II C propeptide level in the synovial fluid as a predictor of radiographic progression in early knee osteoarthritis. Ann Rheum Dis. 2003;62(1):27–32. [PMC free article] [PubMed]
17. Pavelka K, Forejtova S, Olejarova M, et al. Hyaluronic acid levels may have predictive value for the progression of knee osteoarthritis. Osteoarthritis Cartilage. 2004;12(4):277–283. [PubMed]
18. Sharif M, Saxne T, Shepstone L, et al. Relationship between serum cartilage oligomeric matrix protein levels and disease progression in osteoarthritis of the knee joint. Br J Rheumatol. 1995;34(4):306–310. [PubMed]
19. Petersson IF, Boegard T, Svensson B, Heinegard D, Saxne T. Changes in cartilage and bone metabolism identified by serum markers in early osteoarthritis of the knee joint. Br J Rheumatol. 1998;37(1):46–50. [PubMed]
20. Garnero P, Sornay-Rendu E, Arlot M, Christiansen C, Delmas PD. Association between spine disc degeneration and type II collagen degradation in postmenopausal women: the OFELY study. Arthritis Rheum. 2004;50(10):3137–3144. [PubMed]
21. Meulenbelt I, Kloppenburg M, Kroon HM, et al. Urinary CTX-II levels are associated with radiographic subtypes of osteoarthritis in hip, knee, hand, and facet joints in subject with familial osteoarthritis at multiple sites: the GARP study. Ann Rheum Dis. 2006;65(3):360–365. [PMC free article] [PubMed]
22. Goode A, Marshall SW, Renner JB, Carey TS, Kraus VB, Irwin DE, Sturmer T, Jordan JM. Lumbar spine radiographic features and demographic, clinical and concomitant knee, hip and hand OA: The Johnston County Osteoarthritis Project. Arthritis Care and research. 2012 (accepted) [PMC free article] [PubMed]
23. Allen KD, Chen JC, Callahan LF, et al. Associations of occupational tasks with knee and hip osteoarthritis: the Johnston County Osteoarthritis Project. J Rheumatol. 37(4):842–850. [PMC free article] [PubMed]
24. Jordan JM, Helmick CG, Renner JB, et al. Prevalence of knee symptoms and radiographic and symptomatic knee osteoarthritis in African Americans and Caucasians: the Johnston County Osteoarthritis Project. J Rheumatol. 2007;34(1):172–180. [PubMed]
25. Burnett SJHD, Cooper C, Spector TD. A Radiographic Atlas of Osteoarthritis. London: SpringereVerlag; 1994.
26. Kellgren JH. The Epidemiology of Rheumatic Diseases. Annals of the rheumatic diseases. 1964;23:109–122. [PMC free article] [PubMed]
27. Jordan JM, Linder GF, Renner JB, Fryer JG. The impact of arthritis in rural populations. Arthritis care and research. 1995;8(4):242–250. [PubMed]
28. Kraus VB, Jordan JM, Doherty M, et al. The Genetics of Generalized Osteoarthritis (GOGO) study: study design and evaluation of osteoarthritis phenotypes. Osteoarthritis Cartilage. 2007;15(2):120–127. [PubMed]
29. Nelson AE, Chaudhary S, Kraus VB, et al. Whole blood lead levels are associated with biomarkers of joint tissue metabolism in African American and white men and women: The Johnston County Osteoarthritis Project. Environ Res. 2011 [PMC free article] [PubMed]
30. Brant R. Assessing proportionality in the proportional odds model for ordinal logistic regression. Biometrics. 1990;46(4):1171–1178. [PubMed]
31. Ananth CV, Kleinbaum DG. Regression models for ordinal responses: a review of methods and applications. Int J Epidemiol. 1997;26(6):1323–1333. [PubMed]
32. Scott SC, Goldberg MS, Mayo NE. Statistical assessment of ordinal outcomes in comparative studies. J Clin Epidemiol. 1997;50(1):45–55. [PubMed]
33. Peterson B, Harrell FE., Jr Partial Proportional Odds Models for Ordinal Response Variables. Journal of the Royal Statistical Society Series C (Applied Statistics) 1990;39(2):205–217.
34. Rothman K, Greenland S. Modern Epidemiology. 2. Philadelphia, PA: Lippincott Williams & Wilkins; 1998.
35. Hassett G, Hart DJ, Manek NJ, Doyle DV, Spector TD. Risk factors for progression of lumbar spine disc degeneration: the Chingford Study. Arthritis Rheum. 2003;48(11):3112–3117. [PubMed]
36. van Spil WE, DeGroot J, Lems WF, Oostveen JC, Lafeber FP. Serum and urinary biochemical markers for knee and hip-osteoarthritis: a systematic review applying the consensus BIPED criteria. Osteoarthritis and cartilage/OARS, Osteoarthritis Research Society. 2010;18(5):605–612. [PubMed]
37. Cauley JA, Lui LY, Stone KL, et al. Longitudinal study of changes in hip bone mineral density in Caucasian and African-American women. J Am Geriatr Soc. 2005;53(2):183–189. [PubMed]
38. Dragomir AD, Kraus VB, Renner JB, et al. Serum cartilage oligomeric matrix protein and clinical signs and symptoms of potential pre-radiographic hip and knee pathology. Osteoarthritis and cartilage/OARS, Osteoarthritis Research Society. 2002;10(9):687–691. [PubMed]
39. Golightly YM, Marshall SW, Kraus VB, et al. Biomarkers of incident radiographic knee osteoarthritis: Do they vary by chronic knee symptoms? Arthritis and rheumatism. 2011;63(8):2276–2283. [PMC free article] [PubMed]
40. Sturmer T, Raum E, Buchner M, et al. Pain and high sensitivity C reactive protein in patients with chronic low back pain and acute sciatic pain. Ann Rheum Dis. 2005;64(6):921–925. [PMC free article] [PubMed]