Response to the letter to the editor: Computed tomography Hounsfield unit values as a treatment response indicator for spinal metastatic lesions in patients with non-small-cell lung cancer: a retrospective study in Japan

Article information

Asian Spine J. 2025;19(4):677-678

Publication date (electronic) : 2025 August 5

doi : https://doi.org/10.31616/asj.2025.0237.r2

Sho Dohzono

Department of Orthopedic Surgery, Yodogawa Christian Hospital, Osaka, Japan

Corresponding author: Sho Dohzono, Department of Orthopaedic Surgery, Yodogawa Christian Hospital, 1-7-50 Kunijima Higasiyodogawa-ku, Osaka city, Osaka, 533-0024, Japan Tel: +81-6-6632-2250, Fax: +81-6-6320-6308, E-mail: shodohzono@yahoo.co.jp

Received 2025 April 24; Revised 2025 May 3; Accepted 2025 June 27.

Dear Editor,

We appreciate the thoughtful and constructive comments on our article [1], “Computed tomography Hounsfield unit values as a treatment response indicator for spinal metastatic lesions in patients with non-small-cell lung cancer: a retrospective study in Japan.”

Firstly, we agree with your insightful observation regarding survivorship bias. Patients who died within the first 2 months of diagnosis were excluded from our analysis, which may indeed limit generalizability. However, we believe this exclusion was appropriate given the purpose and methodology of our study. At our institution, follow-up computed tomography imaging is routinely conducted only in clinically stable patients. For those with rapid clinical deterioration, such imaging is typically not performed due to limited clinical opportunity for longitudinal assessment. Therefore, incorporating these patients into the Hounsfield units (HU)-based analysis was not feasible and could have compromised the internal validity of the results.

While we acknowledge that this exclusion could introduce bias, our primary objective was to assess the prognostic significance of HU changes in patients for whom follow-up imaging is both clinically applicable and routinely obtained. Future studies should consider including such early mortality cases through sensitivity analyses or alternative approaches to further improve the generalizability of findings.

Secondly, the importance of measurement reliability is a critical point. Although previous studies have investigated intra- and inter-observer variability of HU measurements in the lumbar spine, similar assessments have yet to be performed specifically for metastatic spinal lesions. Future investigations should therefore incorporate reliability metrics such as intraclass correlation coefficients and explore automated or volumetric approaches to enhance reproducibility and objectivity.

Thirdly, the use of a 124 HU cutoff value to define “HU responders” was a practical choice for exploratory stratification, rather than one grounded in biological validation. This approach allowed us to examine patterns in treatment response and survival within our dataset. However, we agree that future research should aim to define and validate clinically meaningful HU thresholds using statistically robust approaches such as receiver operating characteristic curve analysis.

Importantly, even in the field of osteoporosis, where HU has been more extensively studied, a universally accepted threshold has not yet been established. This reflects the variability in imaging conditions and patient populations. Similarly, in metastatic disease, standardization remains a challenge. To address this, larger-scale prospective studies are needed to explore and refine HU thresholds. Ultimately, a comprehensive meta-analysis based on such prospective data will be essential to define evidence-based cutoffs that can inform clinical decision-making.

Fourthly, we acknowledge that key molecular biomarkers such as epidermal growth factor receptor (EGFR)/anaplastic lymphoma kinase (ALK) mutation status and programmed death-ligand 1 (PD-L1) expression were not fully incorporated into our multivariate analysis. EGFR mutation status was indirectly reflected by the inclusion of molecular targeted therapy in the modified Katagiri score. However, ALK mutation status was not used as an independent explanatory variable, as the number of ALK-positive cases in our cohort was insufficient for meaningful analysis.

PD-L1 expression was not assessed in the present study, and we recognize this as an important consideration for future research. While the modified Katagiri score provided a practical prognostic framework incorporating clinical factors such as visceral metastases and performance status, it may not fully capture the molecular complexity of each patient's disease. Future studies should aim to incorporate a more comprehensive range of molecular and clinical variables to improve the accuracy and generalizability of predictive models.

Fifthly, we agree that the retrospective, single-center nature, and relatively small sample size of our study limit its generalizability. Nonetheless, we consider our study to be a preliminary investigation into the potential utility of HU as a prognostic imaging biomarker in patients with spinal metastases. We hope our findings will serve as a foundation for future prospective, multicenter studies with standardized imaging protocols and broader patient cohorts, which will be essential for validation and clinical translation.

In conclusion, we sincerely thank the authors for their constructive and thoughtful critique, which will help refine future work in this important field. We believe that HU measurement is not only a practical tool for evaluating bone density in osteoporosis but also holds promise as an imaging biomarker for monitoring treatment response and predicting survival in patients with spinal metastases.