Assessment of bone mineral density in patients on hemodialysis: diagnostic capabilities of dual-energy X-ray absorptiometry and limitations of FRAX

Full Text

RATIONALE
Chronic kidney disease (CKD) remains a significant medical and social problem in modern healthcare. Renal replacement therapy can increase the life expectancy of patients with end-stage renal failure [1]. Improved survival is accompanied by high-frequency sequences, particularly bone-mineral metabolism disorders, collectively known as CKD-ICD [2, 4, 5]. This category of patients is characterized by a complex combination of deficiency of active vitamin D metabolites, calcium-phosphorus metabolism, secondary hyperparathyroidism, and changes in bone remodeling, ultimately leading to the development of renal osteodystrophy and decreased bone strength [2–5].
According to KDIGO guidelines, bone mineral density (BMD) testing in patients with CKD stages 3–5 is appropriate in cases where the test results can guide the choice of therapeutic strategy [2]. The most common and standardized instrumental method for such measurement is dual-energy X-ray absorptiometry (DXA, DXA), which remains the primary method for quantitatively assessing BMD in clinical practice [6, 7]. For such results, T-scores are used in men aged 50 years and older and postmenopausal women, while in younger patients, it is recommended to use Z-scores [7, 8, 9]. However, in patients with stage 5 CKD receiving program hemodialysis, the standard interpretation of densitometric parameters has a number of reasons due to both CKD-MBD (chronic kidney disease - mineral and bone disorders,
disorders of mineral and bone metabolism in kidney disease), as well as a high degree of concomitant extraskeletal calcifications and degenerative-dystrophic changes [2, 4, 6]. The routine clinical assessment of bone status in this population determines the high prevalence of low-energy changes, which in hemodialysis patients are associated with significant functional factors, increased mortality, and decreased quality of life [9, 10]. Proximal femur fractures are the most significant, as they have the greatest impact on loss of independence and a poor prognosis [9, 10]. Current guidelines and expert opinions indicate that DXA values ​​have prognostic value in patients with end-stage CKD; however, the accuracy of lower zone measurements may vary [2, 6, 11].
Interpretation of lumbar interior bone mineral density data presents a particular challenge. In hemodialysis patients, BMD values ​​in this region may be elevated due to osteophytes, spondylosis, calcification of the abdominal aorta, and other extraskeletal structures within the random zone [6, 11, 12]. These proximal femoral bone parameters, particularly the femoral neck, have been shown to be reliably protected and are significant for bone loss risk management strategies in this population [6, 10, 12–17].
Additional limitations with the FRAX algorithm. Despite its widespread use in the general setting, it has limited validity in patients on hemodialysis because it does not take into account a number of key factors for CKD-MBD, including dialysis duration, parathyroid hormone levels, calcium-phosphorus metabolism disorders, and vitamin D status [10, 12–14]. Therefore, FRAX results in patients with stage 5 CKD should be considered only as auxiliary and interpreted in light of clinical, laboratory, and densitometric data [10, 14]. Thus, for patients with stage 5 chronic kidney disease undergoing hemodialysis, it is necessary to clarify the diagnostic value of DXA in various anatomical sites, primarily the femoral neck and lumbar spine, and to evaluate the role of FRAX in comprehensive risk stratification of bone stability impairment [2, 6, 10, 14].
Study Objective
To evaluate the diagnostic value of dual-energy X-ray absorptiometry in patients with stage 5 chronic kidney disease undergoing hemodialysis and to determine the most informative measurement zone, taking into account the monitoring methodology.

METHODS
STUDY DESIGN
The study had a cross-sectional comparative design. Collection of clinical, laboratory, and densitometric data was planned before the analysis of the results, meaning the study was prospective. The main and control groups were formed at the same clinical site within a predetermined observation period. Consecutive long-term follow-up of participants and repeated measurements were not conducted in this study.
The main group included patients from the chronic hemodialysis unit who met the inclusion criteria, as they became available during the study period. Thus, the sample was consecutive and convenient. The control group consisted of patients from a multidisciplinary hospital without chronic kidney disease and not undergoing program hemodialysis, examined during the same period at the same clinical site. Randomization was not used in group formation. The study did not include blinding, interventions, or assessment of diagnostic accuracy relative to an independent reference outcome. The objective of the study was to comparatively study densitometric parameters in various anatomical zones and analyze their clinical interpretation in patients undergoing hemodialysis.
STUDY CONDITIONS
The study was conducted at the clinical facilities of the Problem Research Laboratory "Diagnostic Research and Minimally Invasive Technologies" of the Federal State Budgetary Educational Institution of Higher Education "Smolensk State Medical University" of the Ministry of Health of the Russian Federation (Regional State Budgetary Healthcare Institution "Clinical Hospital No. 1", Smolensk). The study included 32 patients in the chronic hemodialysis department, constituting the main group (Group 1). Among them were 18 men (56.25%) and 14 women (43.75%), aged 27 to 76 years. All patients were Caucasian. ELIGIBILITY (SELECTION) CRITERIA
• Inclusion criteria were a signed informed consent form, a diagnosis of stage 5 chronic kidney disease, and ongoing hemodialysis treatment.
• Exclusion criteria were refusal to participate in the study, pregnancy, breastfeeding, and the presence of an implanted pacemaker.
• Control group - the control group (Group 2) consisted of 20 patients from a multidisciplinary hospital without chronic kidney disease and not undergoing hemodialysis, as confirmed by clinical diagnosis at admission. This group included 6 men (30.0%) and 14 women (70.0%), aged 22 to 75 years, also Caucasian. Patients in the control group were examined using a similar protocol, followed by a comparative analysis of the obtained data. The control group was used for comparative analysis of densitometric and clinical-demographic parameters.

Description of Eligibility Criteria
The main group included patients who simultaneously met the following criteria: signed informed consent to participate in the study; a diagnosis of stage 5 chronic kidney disease; and ongoing hemodialysis at the time of enrollment. This group was chosen based on the study's objective: to evaluate the diagnostic capabilities of dual-energy X-ray absorptiometry in patients with end-stage chronic kidney disease and severe bone mineral metabolism disorders.
Inclusion criteria included patient refusal to participate further in the study, pregnancy, breastfeeding, and the presence of an implanted pacemaker. Exclusion of pregnant and lactating women was due to general radiation safety requirements for X-ray examinations. The presence of a pacemaker was considered a technical limitation for conducting the study within the established local examination protocol. The control group included patients from a multidisciplinary hospital without chronic kidney disease and without scheduled hemodialysis, as confirmed by clinical diagnosis at admission. The control group was used for comparative analysis of densitometric parameters and was not formed using a matching principle.
Age-specific approaches to densitometry interpretation were determined in advance, prior to the study, based on current clinical guidelines and generally accepted practice for DXA interpretation: the T-score was used in men aged 50 years and older and in postmenopausal women, while the Z-score was used in younger patients [6–9].

Matching of Participants
Matching was not used in this study. The main and control groups were formed independently of each other at the same clinical site during the study period. The planned group ratio was not strictly defined; the actual ratio was 32 patients in the main group and 20 patients in the control group.

DIAGNOSTIC METHODS TESTED
Diagnostic Method
The diagnostic method studied was dual-energy X-ray absorptiometry. All study participants underwent DXA using a DS.Densi Maxi device (Russia) after standard calibration in accordance with the device's operating instructions. The study was conducted under standard clinical conditions at the laboratory. Bone mineral density was assessed at the proximal femur, including the femoral neck and the total hip region, as well as at the lumbar spine (L1–L4). In some cases, forearm measurements were additionally analyzed if relevant data were available.
Threshold values ​​for interpreting DXA results were determined prior to the study and were not derived from the data obtained. The categories of normal, osteopenia, and osteoporosis in patients for whom the T-score was used were defined according to generally accepted WHO thresholds: normal - T ≥ -1.0; osteopenia - from -1.0 to -2.4; osteoporosis - T ≤ -2.5 [7]. In men under 50 years of age and women before menopause, interpretation was performed using the Z-score in accordance with DXA guidelines [6–9].

Reference (control) diagnostic method (if applicable)
The study did not use an independent reference diagnostic method as a standard for assessing the diagnostic accuracy of DXA. The aim of the study was to compare densitometric parameters in the study and control groups, as well as to analyze differences between anatomical measurement zones and their clinical interpretation in patients with stage 5 chronic kidney disease on hemodialysis.

Diagnostic Test Results Recording
Densitometric examination results were assessed using the standard output parameters of the DXA device software, followed by clinical interpretation of the obtained data by the study authors.
The individuals assessing the results of the diagnostic method under study had access to patient clinical information and laboratory data, as the study was conducted in a real-world clinical setting and involved a comprehensive interpretation of densitometric parameters. Independent blinding of clinical data was not performed. Since a reference diagnostic method was not used, cross-blinding between the test and reference test is not applicable in this study.
Uncertain or questionable results were not specifically highlighted; if a correct interpretation for a specific anatomical area was impossible, such data were not included in the corresponding comparative analysis, which was reflected in a change in the number of observations by scan area. Missing data from diagnostic test results were not replaced; the analysis was performed using available data. An analysis of the variability (reproducibility) of DXA results was not performed in this study.

SENSITIVITY ANALYSIS
A sensitivity analysis was not performed in this study. The results were considered descriptive and comparative within the available sample.

STATISTICAL PROCEDURES
Planned Sample Size
A preliminary sample size calculation was not performed at the study planning stage. The sample size was determined by the number of patients who met the inclusion criteria and were available for examination during the designated study period at the clinical site. No special conditions for early termination of the study were envisaged.

Statistical data processing was performed using parametric and nonparametric statistical methods in Excel.
Quantitative parameters for distributions close to normal were presented as the mean and standard deviation (M ± SD); for distributions different from normal, as the median and the first and third quartiles (Me [Q1; Q3]). Categorical variables were presented as the absolute number of observations and percentages.
To compare quantitative parameters between two independent groups, the Student's t-test was used for normal distributions and the Mann-Whitney test for non-normal distributions. Pearson's χ² test was used to compare categorical variables; Fisher's exact test was preferred when the expected number of observations in the contingency table cells was small. Spearman's correlation analysis was performed to assess the relationship between parathyroid hormone levels and densitometric parameters. Differences were considered statistically significant at p < 0.05.
Missing data replacement, data transformation, and special outlier handling were not performed in this study. The analysis was performed using available data; the number of observations was reported separately for the relevant anatomical regions and laboratory parameters. Multivariate regression analysis as a confirmatory step was not included in the final analysis due to the limited sample size and the insufficient number of events classified as "osteoporosis" in the analyzed regions. For this reason, interpretation of the study results is based on descriptive and comparative statistical analysis.

All patients included in the study were examined using a uniform diagnostic algorithm (Fig. 1). Stage 1. Complaints, Medical History, Questionnaire.
During the first stage, complaints, medical history, and questionnaires were collected from patients. Age, gender, anthropometric measurements, duration of renal replacement therapy with hemodialysis, and the presence of comorbidities were assessed. Additionally, clinical data potentially relevant for assessing bone tissue condition and fracture risk were analyzed.
Stage 2. Laboratory Data.
During the second stage, laboratory parameters were analyzed, including creatinine and urea concentrations, estimated glomerular filtration rate, parathyroid hormone (PTH), total calcium, phosphorus, 25(OH) vitamin D, hemoglobin, and ferritin. These parameters were considered as parameters characterizing the severity of mineral and bone metabolism disorders, as well as the general clinical and laboratory status of the patients.

Stage 3. Dual-energy X-ray absorptiometry.
In the third stage, all patients underwent dual-energy X-ray absorptiometry using a DS.Densi Maxi device (Russia). The study was conducted after standard equipment calibration in accordance with the device's operating instructions. Bone mineral density measurements were made at the proximal femur and lumbar spine (L1–L4) in accordance with current clinical guidelines. The T-score was used to interpret the results in men aged 50 years and older and in postmenopausal women; the Z-score was used in younger patients. When analyzing the densitometry results, special attention was paid to comparing the measurements of various anatomical zones in patients with stage 5 chronic kidney disease receiving hemodialysis, taking into account the potential impact of mineral metabolism disorders and bone remodeling characteristics (Fig. 2).

Stage 4. Fracture risk assessment using FRAX.
In the fourth stage, the 10-year risk of major osteoporotic fractures and hip fractures was estimated using the FRAX algorithm. The following parameters were taken into account: region and country of residence, age, gender, weight, height, history of a previous fracture, parental history of a hip fracture, smoking status at the time of assessment, glucocorticoid use, presence of rheumatoid arthritis, secondary osteoporosis, alcohol consumption of 3 or more units per day, as well as femoral neck bone mineral density and the corresponding T-score. FRAX values ​​were used as an additional tool for fracture risk stratification in comparison with densitometric parameters.

RESULTS
In the study group, patients were, on average, older than the control group and differed in gender composition, while anthropometric parameters were comparable between the groups. Laboratory data for the control group were not analyzed and are presented only to characterize the cohort of patients receiving programmatic hemodialysis. These differences should be taken into account when interpreting intergroup densitometric differences (Table 1).
Table 2 presents DXA T-score values ​​for the femoral neck, femur, lumbar spine, and forearm with descriptive statistics. Table 3 presents the distribution of WHO categories. In the study group, T-score values ​​were lower than in the control group, particularly at the femoral neck and overall for the femur; for the lumbar spine, wider variability and a tendency toward "false normalization" were observed, which is methodologically expected in patients on hemodialysis.

In patients in the study group, the most pronounced decrease in the T-score was observed at the femoral neck and total hip. Higher variability in the lumbar spine was observed, which may be due to the influence of degenerative-dystrophic changes and extraskeletal calcifications. Comparison of forearm data between groups is limited due to the extremely small sample size in the control group. Taken together, the results indicate that femoral neck assessment is more informative and that careful interpretation of lumbar spine data is necessary.

In the subgroup of patients for whom densitometric data were interpreted using the Z-score, no Z values ​​≤ -2.0, corresponding to bone mineral density below the expected value for a given age, were recorded. However, the small subgroup size and the limited number of studies in individual anatomical regions preclude this analysis from being considered a standalone source of statistically robust conclusions; therefore, the Z-score results should be assessed as additional descriptive data (Table 3).

Figure 3 shows the distribution of FRAX values ​​in patients in the study and control groups. The columns represent the frequency of values, and the dashed vertical lines represent the medians in each group.

In the study group, FRAX values ​​shifted toward higher values ​​compared to the control group, but a wide overlapping range remained, and absolute risk levels were predominantly low. This is consistent with the limited validity of FRAX in dialysis patients: the algorithm does not take into account dialysis duration or markers such as PTH, phosphorus, or 25-OH vitamin D. Therefore, FRAX should be considered a supplementary indicator, interpreted in conjunction with clinical data and DXA, with priority given to assessing the femoral neck; FRAX alone is inappropriate for making management decisions.

Figure 3 Figure 4 shows a scatter plot of the relationship between the femoral neck T-score and key markers of bone mineral impairment in CKD: parathyroid hormone (PTH) and 25-OH vitamin D. Individual observations, the fitting linear regression line, and the values ​​of the Spearman rank correlation coefficient (ρ), p, and sample size (n) are shown for each graph.
A trend toward an inverse relationship between the femoral neck T-score and PTH levels is noted: ρ=−0.56; p=0.058; n=12 (a decrease in T with higher PTH), which is consistent with the CKD-MBD concept, although statistical significance is borderline with the current sample size. No association with 25-OH vitamin D was detected in these data: ρ=−0.19; p=0.691; n=7. The results highlight the need to interpret laboratory markers in conjunction with DXA and clinical findings, maintaining priority for hip assessment and caution in interpreting lumbar spine parameters; further research in this area with a larger patient population is needed.

DISCUSSION
The results show that in patients with stage 5 chronic kidney disease receiving hemodialysis, densitometric parameters in the proximal femur, particularly the femoral neck, are characterized by a more pronounced decrease compared to the control group, while lumbar spine data demonstrate greater variability and require more cautious interpretation. Thus, in the present study, the femoral regions demonstrated the greatest diagnostic value for detecting decreased bone mineral density in this patient group.
The difference between the proximal femur and lumbar spine parameters deserves special attention. In the study cohort, less pronounced deviations were more often observed for L1–L4 than would be expected in the presence of chronic kidney disease associated with long-term renal replacement therapy. This situation may be due to the influence of degenerative changes in the spine, osteophytes, spondylosis, and extraskeletal calcifications, which can lead to an overestimation of bone mineral density in this anatomical region. Furthermore, in patients with CKD-MBD, structural and metabolic bone changes are mixed, which also complicates the direct interpretation of lumbar spine measurements. Therefore, DXA results in the L1-L4 region in hemodialysis patients should not be considered in isolation, but rather in the context of hip data, laboratory parameters, and overall clinical status.
These findings are generally consistent with published studies demonstrating a high incidence of decreased bone mineral density and an increased risk of fragility fractures, particularly in the hip, in hemodialysis patients. The literature also emphasizes that densitometric assessment of the spine in this patient population may be less reliable due to the high prevalence of degenerative changes and soft tissue calcification. In this context, our results complement existing data, demonstrating that even in real-world clinical settings, differences between the femoral and lumbar spine can have not only quantitative but also interpretative value.
Of particular interest is the analysis of the relationship between densitometric parameters and bone mineral metabolism parameters. In this study, the role of parathyroid hormone as a potential marker of bone remodeling was confirmed by a tendency toward an inverse relationship with femoral neck parameters, which appears pathophysiologically consistent in patients with secondary hyperparathyroidism and chronic kidney disease. However, the lack of consistent statistically significant correlations for a number of parameters precludes the use of individual laboratory markers as an independent alternative to densitometry. It is more likely that they have the greatest clinical value when used as part of a comprehensive assessment that includes DXA, laboratory parameters, and clinical data.
The FRAX score in this study was of an auxiliary nature. Despite its widespread use in the general population, its use in patients on regular hemodialysis has obvious limitations, as the basic model does not account for a number of parameters that are specific to CKD-MBD, including dialysis duration, the severity of calcium-phosphorus metabolism disorders, parathyroid hormone levels, and vitamin D status. For this reason, FRAX results in the study cohort should be interpreted as an additional risk stratification tool rather than as a standalone basis for clinical decisions. When combined with DXA data, particularly at the femoral neck, this score can provide indicative value, but requires careful consideration of the nephrological and metabolic context.
The practical significance of these results is that, when performing densitometric examinations of patients with stage 5 chronic kidney disease undergoing hemodialysis, it is advisable to focus on femoral neck and total hip measurements, as these areas are less susceptible to artifacts associated with degenerative changes in the spine and extraskeletal calcification. Furthermore, interpretation of the results should be based on a standardized examination protocol and take into account the specific bone mineral disturbances characteristic of CKD-MBD. In this sense, densitometry can be considered a useful component of a comprehensive assessment of the risk of decreased bone strength in dialysis patients, but not as the sole diagnostic tool.
This study has several limitations. First of all, it should be noted that the small sample size and the varying number of observations in individual anatomical areas limit the statistical power of the analysis and reduce the robustness of the results.

CONCLUSION
In patients with stage 5 chronic kidney disease undergoing hemodialysis, the most informative densitometric parameters were those of the proximal femur, particularly the femoral neck. Lumbar spine parameters should be interpreted with caution due to the possible influence of degenerative changes and extraskeletal calcification. The FRAX algorithm should only be used as an auxiliary tool. The obtained data justify the feasibility of further studies on larger samples, including clinical outcomes and an expanded analysis of bone mineral metabolism disorders.

AUTHOR'S CONTRIBUTIONS
D.S. Magomedbekova – study design, data analysis and interpretation, manuscript revision at all stages.
E.A. Prolomova – study execution, graphic design, and original text writing.
A.V. Ivanova – patient sample selection and final conclusions.
I.S. Yagubova - study concept, writing the original text.
A. V. Borsukov - scientific supervision, development of study protocols.
D. Yu. Shestakova - writing the original text, final conclusions.

ETHICAL REVIEW
The study was approved by the local ethics committee at the Smolensk Regional State Healthcare Institution "Clinical Hospital No. 1" (extract from meeting minutes No. 18 dated March 11, 2025). Informed consent was obtained from all study participants.
CONSENT FOR PUBLICATION
SOURCES OF FUNDING
None
DISCLOSURE OF INTERESTS
No relationships, activities, or interests over the past 36 months with third parties (individuals or legal entities) whose interests may be affected by the content of the article. ORIGINALITY STATEMENT
This article uses newly created data.
DATA ACCESS
The authors provide full open (unrestricted) access to the data hosted on an external resource (repository).
GENERATIVE ARTIFICIAL INTELLIGENCE
Not used
REVIEW AND PEER-REVIEW
Submitted to the journal's editorial board on their own initiative

About the authors

Diana Saidovna Magomedbekova

Smolensk State Medical University of the Ministry of Health of the Russian Federation

Author for correspondence.
Email: dianamm02@mail.ru
ORCID iD: 0009-0000-6187-8018
SPIN-code: 2427-3476

Student

Russian Federation, 214019, Russia, Smolensk, Krupskaya St. 28

Ekaterina Alekseevna Prolomova

Smolensk State Medical University of the Ministry of Health of the Russian Federation

Email: catherineprolomova@yandex.ru
ORCID iD: 0009-0007-7577-7569
SPIN-code: 1703-6609

Student

Russian Federation, 214019, Russia, Smolensk, Krupskaya St. 28

Anna Valentinovna Ivanova

Smolensk State Medical University of the Ministry of Health of the Russian Federation

Email: annaivanova.ivanova2002@yandex.ru
ORCID iD: 0009-0008-0514-3672
SPIN-code: 1630-9233

Student

Russian Federation, 214019, Russia, Smolensk, Krupskaya St. 28

Irina Sergeevna Yagupova

Smolensk State Medical University of the Ministry of Health of the Russian Federation

Email: Shadowmoon1906@gmail.com
ORCID iD: 0009-0002-0410-0773

Student

Russian Federation, 214019, Russia, Smolensk, Krupskaya St. 28

Alexey Vasilyevich Borsukov

Smolensk State Medical University of the Ministry of Health of the Russian Federation

Email: bor55@yandex.ru
ORCID iD: 0000-0003-4047-7252
SPIN-code: 9412-4149
Scopus Author ID: 7801311680
ResearcherId: AAG-1693-2019

MD, PhD, Professor, Director of the Problem Research Laboratory "Diagnostic Research and Minimally Invasive Technologies"

Russian Federation, 214019, Russia, Smolensk, Krupskaya St. 28

Darya Yuryevna Shestakova

Smolensk State Medical University of the Ministry of Health of the Russian Federation

Email: 92darv@gmail.com
ORCID iD: 0000-0001-5497-1476
SPIN-code: 8489-0188
Scopus Author ID: 58199203500
ResearcherId: ABG-7839-2021

MD, PhD, Senior Researcher, Problem Research Laboratory "Diagnostic Research and Minimally Invasive Technologies"

Russian Federation, 214019, Russia, Smolensk, Krupskaya St. 28

References

Supplementary files