Clinical implications of serum GDF15 propeptide as a biomarker for bone metastasis across common solid tumors: a real-world multi-institutional study

Abstract

Bone metastasis (BM) frequently occurs in various types of cancer, particularly in prostate (PCa), breast (BCa), renal (RCC), and lung cancers (LCa), yet no reliable biomarker has been established. In this study, we evaluated the clinical utility of serum growth differentiation factor 15 propeptide (sGDPP), secreted by both osteoblasts and osteoclasts, as a diagnostic biomarker for BM across common solid tumors. A total of 799 participants were enrolled, including 107 healthy donors, 242 patients with PCa, 113 patients with BCa, 159 patients with RCC, and 178 patients with LCa. Among the cancer patients, 398 had no BM and 294 had BM. The diagnostic performance for BM was compared among conventional biomarkers: alkaline phosphatase (ALP), lactate dehydrogenase (LDH), estimated glomerular filtration rate, osteocalcin, bone-specific alkaline phosphatase, tartrate-resistant acid phosphatase 5b, procollagen type I N-terminal propeptide (PⅠNP), and sGDPP. Among all patients, ALP, LDH, PⅠNP, and sGDPP levels were significantly higher in patients with BM than in those without BM. Particularly, sGDPP showed the highest area under the curve values (0.91 in PCa, 0.80 in BCa, 0.81 in RCC, and 0.51 in LCa). Multivariate analysis showed sGDPP is an independent diagnostic biomarker for BM in PCa, BCa, and RCC (p < 0.01). Additionally, sGDPP in patients with osteoporosis did not significantly differ from that in healthy donors, suggesting that sGDPP increases only when tumor cells seed and proliferate in the bones. Overall, sGDPP demonstrated superior diagnostic performance compared to conventional biomarkers and may complement imaging tests in detecting BM.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40364-026-00982-4.

To the Editor,

The development of bone metastases (BM) is associated with skeletal-related events such as pathological fractures and spinal cord compression, leading to diminished quality of life [1]. Epidemiological data indicate that BM occur in approximately 5.2% of patients with solid tumors [2], with markedly higher incidences reported in advanced cancers including breast (BCa), prostate (PCa), lung (LCa), renal (RCC), and thyroid cancers [3]. Given that over 100,000 patients die annually with BM in the United States alone [4], there is a critical need for reliable and minimally invasive diagnostic tools. Current clinical approaches rely heavily on imaging modalities, including computed tomography and bone scintigraphy, which involve radiation exposure [5]. Although bone turnover markers have been explored, their clinical utility remains limited due to low sensitivity and specificity, as well as susceptibility to confounding factors [6–10].

Our recent work identified plasma growth differentiation factor 15 (GDF15) propeptide (pGDPP) as a promising biomarker for BM [11]. GDPP, the pro-domain of the transforming growth factor-β family member GDF15 [12], is secreted by PCa cells, osteoblasts, and osteoclasts, and promotes BM progression through interactions within the bone microenvironment [11]. Prior studies show that pGDPP correlates more strongly with BM burden than conventional markers, including serum mature GDF15 (mGDF15), despite both molecules being co-secreted during GDF15 processing [11].

Based on these findings, the present study investigates the tumor-agnostic diagnostic potential of serum GDPP (sGDPP) across multiple cancer types with the aim of improving both detection and therapeutic monitoring of BMs.

A total of 799 patients (107 healthy donors, 242 PCa, 113 BCa, 159 RCC and 178 LCa) were included in this study. Among the cancer patients, 398 had no BM and 294 had BM. The diagnostic performance for BM was compared among each tumor marker and following biomarkers: alkaline phosphatase (ALP), lactate dehydrogenase (LDH), estimated glomerular filtration rate, osteocalcin (OC), bone-specific alkaline phosphatase (BAP), tartrate-resistant acid phosphatase 5b (TRACP 5b), procollagen type I N-terminal propeptide (PⅠNP), sGDPP, pGDPP and mGDF15. On analyzing all cancer patients together, the levels of ALP, LDH, sGDPP, pGDPP, mGDF15 and PⅠNP were significantly higher in patients with BM than in patients without BM (Table S1, Fig. 1A, S1A). Additionally, we observed a strong correlation between sGDPP and both pGDPP (R² = 0.88) and mGDF15 (R² = 0.75) (Fig. S1B, S1C), and sGDPP levels tended to increase as renal function declined (R² = 0.50) (Fig. S1D). Because sGDPP demonstrated the highest diagnostic performance for detecting BM (the area under the curve (AUC) = 0.76), with the best cutoff value of 1.1 ng/mL among all biomarkers (Table 1, S2, Fig. 1B, S1E, S1F), we decided to focus our assessment on comparing sGDPP alone with other markers. Importantly, sGDPP levels did not differ significantly between patients with osteoporosis and healthy donors (Fig. S1G), suggesting that sGDPP levels increase when cancer cells metastasize to the bones.

Fig. 1.

Comparison of biomarkers that were higher in patients with bone metastasis than in those without in the overall patient cohort. (A) Serum levels of GDPP, ALP, LDH, PⅠNP and tumor markers for each cancer type in healthy donors (n = 107), cancer patients without BM (n = 398), and cancer patients with BM (n = 294) (** p < 0.01; n.s., not significant). (B) Receiver operating characteristic curves of biomarkers elevated in the BM group for each cancer type. GDPP: growth differentiation factor 15 propeptide, ALP: alkaline phosphatase, LDH: lactate dehydrogenase, PⅠNP: procollagen type I N-terminal propeptide, BM: bone metastasis

Table 1.

Diagnostic utility of serum GDPP at different cutoff values for bone metastasis in cancer patients

Cutoff value (ng/ml)	Sensitivity	Specificity	PPV	NPV
All patients with cancer
0.5	0.99	0.18	0.47	0.96
0.72	0.90	0.39	0.52	0.84
1.0	0.70	0.67	0.61	0.75
1.1	0.66	0.74	0.65	0.75
1.5	0.44	0.86	0.70	0.67
1.68	0.40	0.90	0.75	0.67
Prostate cancer
0.5	1.0	0.20	0.58	1.0
0.96	0.87	0.79	0.82	0.84
1.0	0.81	0.83	0.84	0.80
1.1	0.76	0.87	0.87	0.77
1.5	0.57	0.94	0.91	0.66
Breast cancer
0.5	0.97	0.37	0.64	0.90
0.9	0.62	0.92	0.90	0.68
1.0	0.55	0.92	0.89	0.64
1.1	0.52	0.92	0.88	0.62
1.5	0.30	0.96	0.90	0.54
Renal cell carcinoma
0.5	1.0	0.22	0.32	1.0
1.0	0.79	0.62	0.44	0.89
1.1	0.74	0.72	0.50	0.88
1.27	0.70	0.79	0.56	0.88
1.5	0.60	0.87	0.63	0.86
Lung cancer
0.5	0.98	0.03	0.36	0.75
0.95	0.65	0.43	0.38	0.69
1.0	0.59	0.46	0.37	0.67
1.1	0.54	0.50	0.37	0.67
1.5	0.22	0.74	0.32	0.63

PPV: positive predictive value, NPV: negative predictive value, GDPP: growth differentiation factor 15 propeptide

Next, we assessed the diagnostic performance of sGDPP for detecting BM across tumor types. In PCa, ALP, LDH, sGDPP, BAP, PⅠNP, and prostate-specific antigen were elevated in patients with BM (Table S3, Fig. S2A), with sGDPP demonstrating the highest diagnostic accuracy (AUC = 0.91) and superior sensitivity relative to other biomarkers (Table 1, Table S4-S6, Fig. 1B, S2B-S2D). Similarly, in BCa, ALP, LDH, sGDPP, cancer antigen 15 − 3, and carcinoembryonic antigen (CEA) were increased in patients with BM (Table S7, Fig. 1A, S3A), and sGDPP again showed the best performance (AUC = 0.80) (Table 1, S8-9, Fig. 1B, S3B, S3C). In RCC, elevations in ALP, LDH, sGDPP, and PⅠNP were also observed (Table S10, Fig. 1A, S4A), with sGDPP providing the highest accuracy (AUC = 0.81) and sensitivity (Table 1, S11-12, Fig. 1B, S4B, S4C). In LCa, only CEA was elevated in patients with BM, and no biomarker showed meaningful diagnostic performance (Table 1, S13-15, Fig. 1A and B, S5A, S5B). This limited utility of sGDPP may reflect a lower prevalence of extensive BM (≥ 10 sites) compared with other cancers (Fig. S5C).

In summary, sGDPP demonstrated superior diagnostic performance for BM with conventional bone turnover markers and tumor markers in PCa, BCa, and RCC. The use of sGDPP enable the reduction of radiation-based imaging modalities and facilitate earlier therapeutic intervention, thereby potentially improving patient prognosis.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

BM

Bone metastasis

BCa

Breast cancer

PCa

Prostate cancer

LCa

Lung cancer

RCC

Renal cell carcinoma

GDF15

Growth differentiation factor 15

sGDPP

Serum growth differentiation factor 15 propeptide

pGDPP

Plasma growth differentiation factor 15 propeptide

mGDF15

Mature GDF15

ALP

Alkaline phosphatase

LDH

Lactate dehydrogenase

OC

Osteocalcin

BAP

Bone alkaline phosphatase

TRACP 5b

Tartrate-resistant acid phosphatase 5b

PⅠNP

Procollagen I N-terminal propeptide

AUC

Area under the curve

PSA

Prostate-specific antigen

CEA

Carcinoembryonic antigen

Author contributions

G.Y., T.K., and M.U. conceived the study and designed the experiments; G.Y. and A.M performed the statistical analyses; G.Y., N.M., H.H., J.O., T.N., N.K., M.N., K.N., K.K., K.N., I.S., K.K., K.M., S.S., Y.M., and S.M. collected the patient samples; G.Y. and A.M. analyzed the patient samples; G.Y., A.M., and T.K. prepared the manuscript; and T.K. and M.U. supervised the project. All authors contributed to the critical revision of the manuscript.

Funding

This study was supported by grants from the Japan Society for the Promotion of Science, KAKENHI (20K23002, 21K09396, 24K12436 to G.Y. and 22H04923 (CoBiA) to Y.M.), and from the Japan Society of Clinical Oncology, KFCR grants for young researchers (to G.Y.) and Clinical Research Grant Program (to T.K.).

Data availability

The datasets used and analyzed during the current study is available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Before blood collection, written informed consent was obtained from all participants. All procedures were conducted in accordance with institutional ethical guidelines and approved protocols from the Institutional Review Board of the University of Osaka Hospital (#22101(T1)-11) and Kanagawa Cancer Center Research Institute (2022 − 135).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.