A novel diagnostic model for brucellosis spondylitis using serum PINP and inflammatory factors: a dual-cohort study from Northwest China Authors: Lian Zhou, Guofen Zeng, Shasha Ma, Mingna Li, Shutao Lin, Jing Luo, Lubiao Chen, Chao Wu Published: 2025 ABSTRACT AbstractBackgroundWhile the diagnosis of brucellosis spondylitis (BS) currently relies on imaging, this approach can delay critical intervention. Therefore, identifying novel serum biomarkers is essential to enable earlier diagnosis and prompt antibacterial treatment.MethodsA retrospective analysis was conducted on 78 patients with brucellosis, who were categorized into two groups: those with spondylitis (BS group) and those without (NBS group). We measured plasma levels of the bone turnover markers PINP and CTX-I, along with the inflammatory cytokines TNF-α and IL-1β. Subsequently, univariate and multivariate analyses were performed to identify independent risk factors for BS, which were then incorporated into a logistic regression-based diagnostic model. The performance of this model was evaluated.ResultsComparative analysis revealed significantly elevated plasma PINP levels but decreased IL-1β levels in the BS group relative to the NBS group, whereas no significant intergroup differences were observed in CTX-I or TNF-α levels. A diagnostic model incorporating age, low back pain, ESR, CRP, PINP, and IL-1β was subsequently developed. This model demonstrated strong discriminatory power, with an area under the curve (AUC) of 0.958 (95% CI: 0.915–1.000, P < 0.001) in the derivation cohort. During prospective validation, it maintained an AUC of 0.798 (95% CI: 0.641–0.955), with a specificity of 86.8% and a negative predictive value of 90.4%, despite a sensitivity of 50.0% and a positive predictive value of 46.2%.ConclusionsOur study identified significantly elevated plasma PINP levels in BS patients, forming the basis of a diagnostic model with considerable clinical potential. Nevertheless, further validation through prospective, multi-center studies with larger cohorts is essential to confirm its generalizability.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12879-025-12404-1.Keywords: Brucellosis, Spondylitis, Type I procollagen amino-terminal progenitor peptide, C-terminal propeptide of type I collagen, Diagnosis modeling Introduction Brucellosis, a zoonotic disease caused by Brucella species, poses a significant threat to global public health and the livestock industry [1]. The infection can invade spinal structures, peripheral joints, the meninges, and the endocardium, frequently resulting in severe complications that elevate therapeutic complexity, increase the risk of disability, and raise mortality rates [2]. Of these, brucellosis spondylitis (BS) is one of the most common and diagnostically challenging manifestations. The early diagnostic difficulty is multifaceted. First, the gold-standard method—biopsy and culture—has low sensitivity, with a reported positivity rate of only approximately 32% [3]. Second, clinical differentiation from other spinal infections remains challenging. For example, tuberculous spondylitis (TS) and BS present with highly similar clinical features, such as fever, back pain and the increase of inflammatory indicators, making them difficult to distinguish in practice [4]. In fact, current diagnostic paradigms of BS primarily rely on magnetic resonance imaging (MRI) findings [5–7]. However, MRI has notable limitations, particularly in early disease stages when radiographic changes may be nonspecific or absent [8]. Although the droplet digital PCR (ddPCR) has recently been introduced for diagnosing BS, its dependency on biopsy specimens—which are invasive and often difficult to obtain—further complicates its clinical application [9]. Given these constraints, there is an urgent need to develop alternative methods for the early diagnosis of BS. Osteoarticular brucellosis is pathophysiologically characterized by an imbalance between excessive osteoclast-mediated bone resorption and insufficient osteoblast-driven bone formation [10–12]. In recent years, several biochemical markers have been increasingly recognized for providing valuable insights into the dynamics of bone remodeling [13, 14]. For instance, the N-terminal propeptide of type I procollagen (PINP), a byproduct of collagen synthesis, exhibits high specificity and sensitivity in quantifying bone formation activity, with serum levels that strongly correlate with histomorphometric parameters. In contrast, the C-telopeptide of type I collagen (CTX-I), a degradation product of bone collagen, serves as the most specific and sensitive biomarker for assessing bone resorption intensity [15, 16]. Additionally, pro-inflammatory cytokines also play a destructive role. Tumor necrosis factor-alpha (TNF-α) directly stimulates osteoclast differentiation and bone degradation [17]. Interleukin-1β (IL-1β) also contributes by upregulating various matrix metalloproteinases (MMPs 2, 3, 4, 5, 9, 13) and the Receptor Activator of Nuclear Factor-κB Ligand (RANKL) expression, thereby promoting bone and cartilage degradation—a key mechanism in early osteoarthritis [18]. Beyond these direct effects, a broader cytokine cascade—involving interferon-gamma (IFN-γ), and interleukins such as IL-4, IL-6, IL-10, and IL-17—has been implicated in modulating the overall progression and severity of brucellosis, including the Brucellosis arthritis [19]. Therefore, the present study was designed to address the challenge of early BS diagnosis by developing and validating a diagnostic model that leverages a panel of serum biomarkers, including PINP, CTX-1, and specific inflammatory mediators. Materials and methods Design and participants This study employed a dual-phase design consisting of a retrospective analysis and a prospective validation cohort. Participants were consecutively recruited from the Department of Infectious Diseases of the First People’s Hospital of Kashi, Xinjiang, China. They were stratified into two groups: the brucellosis spondylitis (BS) group and the brucellosis without spondylitis (NBS) group. A total of 78 patients diagnosed with brucellosis between October 2020 and December 2021 were included in this retrospective analysis (BS = 18, NBS = 60), while an additional 65 patients enrolled between January and December 2022 formed the prospective validation cohort (BS = 12, NBS = 53). Additionally, 28 healthy individuals with no clinical evidence of disease served as healthy controls (HC). The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Ethics Committee of the First People’s Hospital of Kashi ([2022] Fast Review Research No. 06). Written informed consent was obtained from all participants. Diagnostic and exclusion criteria Brucellosis diagnosis was based on the Chinese National Diagnostic Criteria for Brucellosis (WS269-2019) [20], which includes: Epidemiological exposure history. Epidemiological exposure history. Characteristic clinical manifestations, such as fever, fatigue, headache, hyperhidrosis, joint and muscle pain. Characteristic clinical manifestations, such as fever, fatigue, headache, hyperhidrosis, joint and muscle pain. Serological confirmation of a positive Rose Bengal plate agglutination test or a standard agglutination test titer ≥ 1:100 or Brucella spp. isolated from blood culture. Serological confirmation of a positive Rose Bengal plate agglutination test or a standard agglutination test titer ≥ 1:100 or Brucella spp. isolated from blood culture. The diagnostic criteria for BS are based on MRI findings [21, 22], which included the presence of (1) vertebral body abnormalities, (2) intervertebral disc changes, (3) spondylolisthesis, (4) ligamentous involvement, (5) inflammatory edema, and (6) paravertebral abscess. Exclusion criteria encompassed: (1) co-existence of active tuberculosis or acquired immunodeficiency syndrome; (2) concurrent presence of fractures of other causes, spinal tuberculosis, bone tumors or other osteoarticular diseases; (3) history of recent orthopedic surgery; (4) presence of renal dysfunction, liver dysfunction, or parathyroid disease; (5) postmenopausal women receiving hormone therapy or patients diagnosed with osteoporosis; (6) cases with serious lack of clinical diagnostic and treatment data, and (7) women with a positive pregnancy test or those actively breastfeeding. Sample processing Fasting venous blood samples were collected on the day of admission, prior to the initiation of antimicrobial therapy. All whole blood specimens were centrifuged at 1,800 g for 10 min, and the resulting plasma samples were carefully collected and stored at -80 °C until batch analysis. Quantitative measurements of PINP, CTX-I, TNF-α, and IL-1β were performed using commercial enzyme-linked immunosorbent assay (ELISA) kits (Bio-swamp, Wuhan, China) following the manufacturer’s protocols. All samples were analyzed in duplicate to ensure accuracy and reproducibility. Statistical analysis Statistical analyses were performed using IBM SPSS Statistics for Windows, version 25.0. Data are presented as means ± standard deviation (SD), median, or rate, depending on the distribution and nature of each variable. Comparisons between the BS and NBS groups were conducted using the independent samples t-test, the non-parametric Mann–Whitney U rank-sum test, the chi-square test, or Fisher’s exact test, as appropriate. To identify potential risk factors for BS, univariate and multivariate logistic regression analyses were conducted. The receiver operating characteristic (ROC) curve was used to determine the cutoff value for diagnosing BS, and ROC curves were generated using GraphPad Prism version 9.0. A two-tailed p-value of 0.05 was considered statistically significant. Results Demographic and clinical characteristics The demographic data, clinical symptoms, laboratory findings, and vertebral involvement patterns are summarized in the Supplementary Materials. The BS group was significantly older than the NBS group (50.44 ± 9.65 years vs. 38.32 ± 11.42 years, P < 0.001) and had longer hospital stays (15.56 days vs. 12.30 days, P = 0.017). Additionally, a significantly higher proportion of patients with BS experienced back pain compared to the NBS group (83.30% vs. 40.00%, P = 0.001) (Supplementary Table 1). No significant differences were observed between the two groups regarding sex distribution, epidemiological exposure history, clinical stage, fever, nausea, weight loss, night sweats, headache, or fatigue. In terms of laboratory parameters, the BS group had significantly higher levels of white blood cells, C-reactive protein (CRP), and erythrocyte sedimentation rate (ESR) compared to the NBS group (Supplementary Table 2). However, there were no significant differences between the two groups in procalcitonin (PCT), IL-6, liver and kidney function tests, brucellosis serological results, or blood culture positivity rates. MRI findings revealed that multiple vertebrae were involved in 66.67% of BS cases, with the lumbar vertebrae being the most frequently affected region (72.2%) (Fig. 1) (Supplementary Table 3). Magnetic resonance imagings of a 40-year-old male patient (A-B) and a 51-year-old male patient (C-D) with BS: low signal intensity on T1WI (A) and uneven high intensity on T2WI (B) indicating L4-L5 vertebral bone destruction; low signal intensity on T1WI (C) and uneven high signal intensity on T2WI (D) indicating vertebral bone destruction (L1-L5), insect erosion at vertebral body edge (L1-L4), and narrowing intervertebral space (L1/L2, L2/L3 and L4/L5) Expression of plasma biomarkers Plasma levels of PINP, CTX-I, TNF-α, and IL-1β were measured in three groups (Supplementary Table 4). Compared with the NBS and HC groups, PINP levels were significantly elevated in the BS group. Although CTX-I levels were elevated in both the NBS and BS groups, the difference between these two groups was not statistically significant. Similarly, TNF-α levels did not differ significantly among the three groups. In contrast, IL-1β levels were substantially lower in the BS group compared to the NBS. (Fig. 2) Expression levels of PINP (A), CTX-I (B), TNF-α(C) and IL-1β(D) in plasma from the healthy controls (HC), brucellosis patients without (NBS) or with spondylitis (BS) Risk factors for BS Following the comparison of clinical parameters and plasma biomarkers between the BS and NBS groups, univariate logistic regression analysis identified older age, low back pain, elevated CRP, elevated ESR, elevated PINP, and decreased IL-1β levels as risk factors for BS (Table 1). However, in the multivariate analysis, only elevated PINP remained an independent risk factor for BS (95% CI: 1.001–1.020; P = 0.032). Other variables, including age, low back pain, CRP, ESR, and IL-1β, were not independently associated with BS. Table 1Analysis of risk factors for brucellosis spondylitis in retrospective cohortParametersUnivariate analysisOR (95%CI)P-valueMultivariate analysisOR (95%CI)P-valueAge (years)1.101 (1.041–1.163) 0.001 1.106 (0.993–1.233)0.067Low back pain7.500 (1.958–28.726) 0.003 2.158 (0.152–30.721)0.570CRP (mg/L)1.021 (1.003–1.038) 0.019 1.014 (0.980–1.048)0.428ESR (mm/H)1.031 (1.008–1.055) 0.009 0.995 (0.958–1.033)0.799PINP (ng/ml)1.014 (1.008–1.019) < 0.001 1.010 (1.001–1.020) 0.032 IL-1β (pg/ml)0.990 (0.985–0.996) 0.001 0.991 (0.980–1.001)0.088 Analysis of risk factors for brucellosis spondylitis in retrospective cohort Establishment of a diagnostic model for BS The diagnostic performances of the six identified risk factors for BS are summarized in Table 2. The optimal cutoff values for each risk factor were determined using ROC curve analysis (Fig. 3). Based on these data, all six risk factors were incorporated into a diagnostic model, represented by the following equation: Table 2Univariate diagnostic performance for BS in retrospective cohortParametersCut OffAUC (95%CI)SensitivitySpecificityP-valueAge (years)39.500.782 (0.671–0.983)0.8890.650 < 0.001 Low back painYes/No0.717 (0.588–0.845)0.8330.400 0.006 CRP (mg/L)8.8550.716 (0.592–0.841)0.9440.448 0.006 ESR (mm/H)25.500.772 (0.657–0.888)0.7650.698 0.001 PINP (ng/ml)127.900.870 (0.757–0.984)0.8890.900 < 0.001 IL-1β (pg/ml)29.230.779 (0.653–0.905)0.9830.017 < 0.001 Univariate diagnostic performance for BS in retrospective cohort ROC curves of age (A), low back pain (B), CRP (C), ESR (D), PINP (E) and IL-1β (F) for diagnosis of brucellosis spondylitis in retrospective analysis Where π represents the probability of spondylitis occurring in patients with brucellosis, X1, X3, X4, X5, and X6 correspond to the patient’s age at disease onset and the levels of CRP, ESR, PINP, and IL-1β, respectively. X2 is assigned a value of 1 if the patient presents with low back and zero if they do not. The area under the curve (AUC) of the diagnostic model was 0.958 (95% CI: 0.915–1.000, P < 0.001) (Fig. 4), with a sensitivity of 88.2%, specificity of 90.6%, positive predictive value (PPV) of 72.2%, and negative predictive (NPV) value of 83.3%. This model outperformed the predictive accuracy of individual risk factors. This model was further validated in a prospective cohort study. ROC curve of the model for diagnosis of brucellosis spondylitis in retrospective analysis Validation of the diagnosis model A total of 65 patients were enrolled in this prospective study, including 12 patients in the BS group and 53 in the NBS group. Consistent with the findings of the retrospective study, BS cases were mainly characterized by multilevel vertebral involvement, with the lumbar vertebrae being the most frequently affected, followed by the lumbosacral vertebrae. Detailed clinical data of the patients are provided in the supplementary materials (Supplementary Tables 5–7). Using spinal MRI as the gold standard for BS diagnosis, the model demonstrated an AUC of 0.798 (95% CI: 0.641–0.955), with a sensitivity of 50.0%, specificity of 86.8%, PPV of 46.2%, and NPV of 90.4%. Discussion In China, brucellosis outbreaks exhibit a distinct geographic concentration, with disproportionately high prevalence in the Northwestern and Northeastern regions. A prime example is Xinjiang, where the incidence rate substantially exceeds the national average [23]. In terms of symptoms, Brucella species can invade the human body via the respiratory tract, digestive tract, or skin, leading to systemic infections characterized by fever and multi-organ damage [24]. Brucellosis often leads to severe complications, with osteoarticular involvement being the most prevalent, occurring in up to 77% of cases and primarily affecting large joints [25]. The lumbar spine is particularly susceptible to such involvement, posing considerable challenges in clinical management [25, 26]. Consistent with this, our data clearly demonstrated that patients with spondylitis require significantly longer hospital stays than those with uncomplicated brucellosis, highlighting the urgent need for early and accurate clinical differentiation. In fact, BS is often misdiagnosed or overlooked, largely due to symptom overlap with uncomplicated brucellosis in the early stages. Our findings confirm that both conditions share common systemic symptoms, such as fever, fatigue, poor appetite, and night sweats. However, low back pain was significantly more prevalent in the BS group, underscoring its value as a key clinical indicator. Nevertheless, the diagnostic performance of back pain alone showed a sensitivity of 0.833 but a specificity of only 0.400 in our study, indicating that it should be considered suggestive rather than pathognomonic for BS. Meanwhile, MRI, while the preferred imaging modality for diagnosing BS and differentiating it from conditions like tuberculous spondylitis [27–29], has notable limitations. It offers limited utility for detecting early-stage BS, and its high cost restricts its suitability as a routine screening tool [30]. Biomarkers play a crucial role in infectious diseases, with applications spanning diagnosis [31, 32], prognosis assessment [33], and evaluation of treatment response [34]. In the context of Brucellosis, several studies have explored the diagnostic potential of inflammatory cytokines. While Togan et al. [35] reported elevated CRP levels in brucellosis patients compared to healthy controls, an increased ESR has been specifically associated with bone and joint involvement [36, 37]. Extending beyond individual markers, a broader panel of biomarkers—encompassing neutrophil counts, neutrophil-to-lymphocyte ratio (NLR), monocyte-to-lymphocyte ratio (MLR), ESR, and CRP—has been reported to be collectively elevated in Brucellosis patients with osteoarticular or specific organ involvement compared to control groups. Building upon prior evidence, we extended these observations to brucellosis spondylitis and identified a significant elevation in total white blood cell and neutrophil counts relative to brucellosis patients without spinal involvement. Notably, both CRP and ESR were significantly elevated in the BS group and were identified as risk factors for this condition. Our evaluation of their diagnostic performance revealed high sensitivity (CRP: 0.944; ESR: 0.765) but comparatively low specificity (CRP: 0.448; ESR: 0.698). The importance of these markers was further underscored by their contribution to the diagnostic model. Specifically, the removal of CRP led to a decline in model sensitivity from 0.882 to 0.722 in the retrospective cohort and from 0.500 to 0.416 in the prospective cohort. A similar reduction in sensitivity was observed upon the exclusion of ESR (retrospective: 0.882 to 0.778; prospective: 0.500 to 0.416), confirming that both markers are integral to the model’s detection capability. These findings confirm that CRP and ESR are essential for preserving the sensitivity of the diagnostic model. Given their role as readily accessible and cost-effective clinical tests, we recommend emphasizing both markers in the diagnostic evaluation of BS to enhance early detection. Unlike in most bacterial infections, PCT levels are typically not elevated in brucellosis [38]. Consistent with this understanding, our findings demonstrate a negligible role for PCT in the diagnosis of BS. As a pivotal pro-inflammatory cytokine, IL-1β acts as a key driver in the progression of osteoarthritis and the development of osteoporosis, primarily by promoting cartilage degradation and disrupting bone remodeling [39]. F. Apostolou et al. [40] longitudinally measured pro-inflammatory cytokines (IL-1β, IL-6, TNF-α) in 28 patients with simple acute Brucella infection at admission and after 4 months of treatment. Levels of all three cytokines were significantly elevated at admission compared to healthy controls. Following successful treatment, IL-1β and IL-6 decreased significantly, whereas TNF-α levels remained persistently elevated. However, our results indicated that serum IL-1β levels were significantly lower in the BS group than in the NBS group. This discrepancy could be attributed to differences in study populations. While F. Apostolou’s study primarily included cases of simple acute brucellosis without major complications, our study specifically compared BS against NBS, focusing on a pathological process dominated by localized bone remodeling. Despite its known role in inflammation, no study has systematically compared IL-1β levels across brucellosis complications. Therefore, evaluating its specific diagnostic utility for BS must be a priority for future research. In addition to inflammatory cytokines, our model also incorporated key aspects of the bone remodeling process. Previous studies have documented bone loss in patients with BS [10–12]. While several biomarkers, such as pyridinoline, type II collagen cleavage peptide, and osteocalcin, have been associated with brucellosis osteoarthritis [41, 42], their diagnostic specificity remains limited. This lack of specificity underscores the need for more reliable biomarkers to enhance clinical utility. In clinical practice, PINP and CTX-I are considered the preferred serum biomarkers for assessing bone turnover, with PINP reflecting bone formation and CTX-I indicating bone resorption [43]. Current research on PINP and CTX-I mainly focuses on osteoarthritis, fracture healing, osteoporosis, and tumor-related bone metastases [44–46]. In our study, PINP but not CTX-I levels were significantly elevated and identified as an independent risk factor for BS, underscoring its potential diagnostic value. This finding is consistent with a body of evidence from osteoarthritis research, reinforcing the validity of our conclusion. Berry et al. [47] studied hand and knee arthritis in patients with obesity and found that baseline leptin levels were associated with increased bone formation biomarkers (osteocalcin and PINP) over two years. Similarly, Kumm et al. demonstrated that PINP not only served as a diagnostic biomarker for knee osteoarthritis but also predicted disease progression, particularly progressive osteophytosis [48]. Current research on predictive models for human brucellosis has incorporated various data types, ranging from routine laboratory parameters like full blood count analyzed with machine learning [49] to cutting-edge proteomic approaches [50]. However, the development of a diagnostic model that integrates inflammatory factors with bone remodeling indicators for BS has not yet been reported. In this study, we developed a diagnostic model incorporating PINP, age, low back pain, CRP, ESR, and IL-1β. While prospective validation revealed lower sensitivity and positive predictive value, the model maintained high specificity and negative predictive value, confirming its strong diagnostic utility. Given that early imaging can often yield negative results, our model shows significant clinical promise by aiding in diagnosis during this critical window. This study has several limitations. The generalizability of our findings is constrained by its single-center design and relatively small sample size, which may introduce selection bias and limit the representation of the broader BS patient population. Additionally, the use of both retrospective and prospective data, while beneficial for initial validation, may lead to inconsistencies in data collection and presentation. Potential confounding factors, such as variations in disease stage and prior treatments, were not fully explored. Future large-scale, multi-center studies are needed to validate these results. Conclusions In conclusion, this study not only identifies elevated plasma PINP as a robust biomarker associated with BS but also validates a diagnostic model with significant clinical potential. These findings suggest that aberrant bone turnover plays a central role in BS pathogenesis. To advance this work, future efforts should focus on: (1) external validation of the model in large-scale, multi-center cohorts to assess its generalizability; (2) elucidating the precise molecular mechanisms linking Brucella infection to PINP upregulation and bone remodeling; and (3) exploring the integration of this model with advanced imaging techniques to further improve diagnostic accuracy. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements We gratefully acknowledge the invaluable assistance provided by Aimin Xu, the director of laboratory, the First People’s Hospital of Kashi. Additionally, we thank all study participants. Abbreviations Area under the curve Brucellosis Spondylitis C-reactive protein C-terminal propeptide of type I collagen Enzyme-linked immunosorbent assay Erythrocyte sedimentation rate Healthy controls Interferon-gamma Interleukin Lymphocyte ratio Monocyte-lymphocyte ratio Matrix metalloproteinases Magnetic Resonance Imaging Neutrophil-Lymphocyte ratio Negative predictive Osteoarticular brucellosis Odds ratio Procalcitonin N-terminal propeptide of type I procollagen Positive predictive value Receptor Activator of Nuclear Factor-κ B Ligand Receiver operating characteristic Standard deviation T1-weighted imaging T2-weighted imaging Tumor necrosis factor Upper limit of normal Author contributions LZ, GFZ, CW and LBC designed the work; CW and JL assisted in patient enrollment and specimen collection; LZ and GFZ collected serum samples and tested serum biomarkers, analyzed data, drafted the manuscript; SSM, MNL and STL collected the data. CW and LBC provided the funding. All authors reviewed the manuscript. Funding This work was supported by grants from Science and Technology Plan Project in Kashi Prefecture (grant number KS20222024). Data availability The datasets used and/or analyzed during the current study are available from the corresponding author or the first author on reasonable request. Declarations Ethics approval and consent to participate The study was approved by the Research Ethics Committee of the First People’s Hospital of Kashi. All procedures performed in the study were in accordance with the 1964 Declaration of Helsinki and its subsequent revisions or similar ethical standards and the ethical requirements of the institutional research ethics committee. Informed consents were obtained from all participants. Consent for publication Written informed consents for clinical and personal details were obtained from all subjects involved in the study. Competing interests The authors declare no competing interests. Footnotes Co-corresponding authors: Chao Wu Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Lian Zhou and Guofen Zeng contributed equally to this work. Contributor Information Lubiao Chen, Email: chenlub@mail.sysu.edu.cn. Chao Wu, Email: 3216233205@qq.com. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author or the first author on reasonable request.