Response to “Letter to the Editor Regarding “Streamlining the MRI-Derived Vertebral Bone Quality Score for Preoperative Osteoporosis and Osteopenia Screening”

Abstract

We thank the authors for their thoughtful and constructive comments on our simplified MRI-derived vertebral bone quality score using L1/L2 measurements VBQ_L1/2. We agree with the authors that multicenter studies incorporating heterogeneous scanners and acquisition protocols are necessary before absolute VBQ cutoff values can be broadly applied. Our study was designed to assess whether VBQ measurement could be simplified from L1–4 to L1/L2 without compromising performance, rather than to establish a universal threshold. VBQ_L1/2 showed excellent agreement with VBQ_L1-4 and comparable discrimination for osteoporosis and impaired bone quality. We therefore consider VBQ an opportunistic pre-screening tool rather than a substitute for DXA or quantitative CT. Given the high burden of undiagnosed or untreated osteoporosis in lumbar fusion patients and the limitations of general-population screening criteria, spine-specific strategies may benefit from combining clinical risk factors with opportunistic imaging biomarkers. Ongoing work on inter-scanner reliability and automated MRI analysis may facilitate the larger multicenter validation studies required to validate or refute the clinical utility of VBQ.

Dear Editor,

We thank the authors for their thoughtful and constructive comments ¹ regarding our study, “Streamlining the MRI-Derived Vertebral Bone Quality Score for Preoperative Osteoporosis and Osteopenia Screening”. ² We appreciate their recognition of the potential clinical utility of a simplified vertebral bone quality (VBQ) measurement and agree that they raise an important consideration regarding the transportability of absolute VBQ cutoff values. Standardization and external validation are essential before absolute VBQ cutoff values can be broadly implemented.

Liu et al. demonstrated significant differences in VBQ according to magnetic field strength and bandwidth, while Lin et al. reported substantially different osteoporosis discrimination between 1.5T and 3.0T MRI.^3,4 These findings indicate that normalization to cerebrospinal fluid does not completely eliminate scanner-related variability and that absolute VBQ thresholds should not presently be assumed to be universally transferable. We share the authors’ view that this represents an important area for further investigation, ideally through multicenter collaboration incorporating heterogeneous scanners and acquisition protocols.

Importantly, the primary objective of our study was not to establish a universal VBQ_L1/2 cutoff, but to determine whether reducing the conventional L1–4 measurement to L1 and L2 could simplify the methodology without compromising its performance. Both VBQ_L1/2 and VBQ _L1-4 were obtained from the same MRI examination in each patient and were therefore exposed to the same scanner and acquisition conditions. We observed excellent agreement between both approaches (ICC 0.988) and comparable discrimination for osteoporosis and impaired bone quality. Thus, our principal finding concerning simplification of the measurement rather than generalizability of the derived threshold. Our original conclusion explicitly called for external validation and cutoff refinement. ²

We also agree that the specificity of 45% associated with the Youden-derived cutoff of 2.30 requires consideration. We propose VBQ_L1/2 as an opportunistic pre-screening tool rather than a diagnostic substitute for DXA or quantitative CT (qCT). In this context, prioritizing sensitivity may be reasonable if a positive result triggers definitive bone mineral density (BMD) assessment rather than treatment itself. Nevertheless, the false-positive burden and downstream resource utilization need to be evaluated prospectively.

Low BMD is associated with mechanical complications after lumbar fusion surgery (LFS), including cage subsidence, screw loosening, pseudarthrosis, and junctional complications, in addition to its general risk of osteoporotic fractures.^5,6 These procedure-specific risks suggest that screening strategies for LFS patients may need to differ from those developed for the general population. Indeed, general-population osteoporosis screening recommendations demonstrated sensitivities of only 78% in women and 58% in men prior to LFS recently. ⁷ In prior work, we demonstrated that almost half of patients with osteoporosis were untreated prior to LFS, with 80% of untreated cases in previously undiagnosed patients. ⁸ In a subsequent analysis of undiagnosed low BMD prior to LFS, 12% of the entire cohort had previously undiagnosed osteoporosis, representing 37% of all osteoporosis cases. ⁹ These findings emphasize both the need for preoperative bone-health assessment and the difficulty of determining whom to screen solely using conventional criteria alone.

Accordingly, opportunistic imaging markers may complement clinical screening algorithms. Beyond VBQ, abdominal aortic calcification on routinely available lumbar radiographs has been associated with impaired fusion after LFS ¹⁰ and with lower BMD as well as alterations in bone microarchitecture and extracellular matrix characteristics independently of BMD. ¹¹ Rather than relying on any single surrogate marker, future spine surgery-specific screening strategies may therefore integrate patient- and procedure-specific risk factors with routinely available imaging to identify patients requiring definitive BMD assessment prior to LFS.

Finally, we fully agree with the authors that larger and more heterogeneous validation studies are required before VBQ can assume such a role. We are currently investigating inter-scanner reliability and comparability of VBQ and developing tools for automated MRI biomarker analysis. The feasibility of automated VBQ assessment has recently been demonstrated using deep-learning approaches. ¹² Automation could substantially reduce measurement burden and enable analysis of large multicenter datasets across different scanners and acquisition protocols. Such studies will ultimately be necessary to validate or falsify the clinical utility of VBQ and determine whether standardized thresholds, scanner-specific calibration, or alternative approaches are required.

We therefore view VBQ_L1/2 as a methodological simplification that may facilitate these next steps, rather than as a validated diagnostic. We appreciate the authors’ thoughtful critique and share their interest in addressing these remaining questions. We hope that continued validation through multicenter collaboration will help establish how VBQ can most reliably be incorporated into opportunistic bone-health assessment in spine surgery.

Sincerely,

The Authors

ORCID iDs

Paul Köhli https://orcid.org/0000-0001-7473-6553

Marco D. Burkhard https://orcid.org/0000-0003-1501-1952

Alexander P. Hughes https://orcid.org/0000-0001-7293-9672