Modified criteria for identifying elevated bone mass

Abstract

Summary

This study used modified criteria to exclude osteoporosis and osteopenia from elevated bone mass (EBM) to avoid a diagnostic paradox and demonstrated that diffuse idiopathic skeletal hyperostosis (DISH), severe lumbar degenerative disease (LDD), elevated body mass index (BMI), and diabetes are key risk factors for super EBM (SEBM).

Purpose

The current criteria classify individuals with both EBM and osteopenia/osteoporosis. To avoid this contradiction, we used a modified criterion to exclude osteoporosis or osteopenia from EBM and to examine the clinical impact and risk factors of EBM.

Methods

In this retrospective study, bone mineral density (BMD) data from participants aged 18 years or older collected at a tertiary hospital between 2021 and 2023 were analyzed. Inclusion criteria were complete Z-scores and T-scores for the lumbar spine, femoral neck, and total hip. Participants with a history of lumbar spine surgery or hip arthroplasty prior to BMD measurement were excluded. Normal bone mass (NBM) was defined as T-scores ≥ –1.0 at all sites and Z-scores < 2.5 at any site. EBM was defined as NBM plus Z-scores ≥ 2.5 at any site. EBM was further subdivided into moderate EBM (MEBM) with Z-scores between 2.5 and 4.0, and SEBM with Z-scores ≥ 4.0. Demographic data, medical history, and comorbidities were collected and analyzed.

Results

Of the 33,479 eligible participants, 1,472 (4.4%) were identified with EBM. The EBM group was divided into 1,267 (3.8%) with MEBM and 205 (0.6%) with SEBM. SEBM group had a significantly higher proportion of men and a higher BMI than the MEBM group (p < 0.001). In addition, SEBM group had a higher prevalence of conditions such as DISH, ankylosing spondylitis (AS), LDD, and diabetes (p < 0.001). Different correlation trends between BMI, T-scores, and Z-scores were observed in the SEBM and MEBM groups. Chi-squared Automatic Interaction Detection (CHAID) analysis identified DISH as the strongest predictor of SEBM, while severe LDD, increased BMI, and diabetes were contributing factors for SEBM.

Conclusions

Using the modified criteria for EBM, which exclude cases of osteoporosis and osteopenia, SEBM has a higher rate of comorbidities compared to MEBM. The presence of DISH, higher severity of LDD, increased BMI, and diabetes were identified as strong risk factor of SEBM.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00198-025-07551-9.

Introduction

The World Health Organization (WHO) diagnostic criteria for osteoporosis and osteopenia are generally accepted and used by medical professionals and the public [1]. However, the WHO definition of normal bone mass (NBM) (T-score ≥ −1.0) is less recognized and may be misleading. For example, certain diseases such as osteopetrosis or otosclerosis may have T-scores within the WHO-defined NBM range but actually have abnormally elevated bone mass (EBM) [2]. In 2002, the Osteoporosis Society of Canada classified NBM as T values between −1.0 and + 2.5 and emphasized that EBM exceeding certain thresholds does not always indicate normal bone health [3].

Currently, the term EBM [4] can also be referred to interchangeably as high bone mass (HBM) [5, 6], high bone mineral density (HBMD) [7, 8] or high bone density (HBD) [2, 9]. However, the cut-off values for defining EBM vary between the limited studies and are not standardized. The cut-off values are based on bone mineral density (BMD) measured by dual-energy X-ray absorptiometry (DEXA). In 2005, Michael Whyte proposed using a Z-score ≥ + 2.5 as a cut-off value to define abnormal BMD [2]. This cut-off value was adopted in a study on the genetic etiology of EBM, in which a Z-score ≥ + 2.5 at least two skeletal sites was used to classify EBM [10]. A study from Canada defined EBM by a T-score ≥ + 2.5 or a Z-score ≥ + 2.0 [7]. Studies from France used T-scores or Z-scores ≥ + 4 for either the lumbar spine, neck or hip to define EBM [4, 6, 8]. A study from the UK recommended complex cut-off values, lumbar spine 1 (L1) Z-score ≥ + 3.2 plus total hip Z-score ≥ + 1.2, or total hip Z-score ≥ + 3.2 [11]. From these studies, it appears that the Z-score or T-score above + 2.5 or + 4.0 may be the two commonly used cut-off values for EBM.

However, there is a potential conflict between the WHO diagnostic criteria and the EBM definition, especially when considering the primary measurement sites of DEXA, which include the femoral neck, lumbar spine, and total hip. It is possible for a patient to simultaneously have a lumbar spine Z-score greater than + 2.5, indicating EBM, and a total hip T-score less than −2.5 or −1.0, indicating osteoporosis or osteopenia [4, 6–8]. This dual diagnosis raises important questions about the validity and applicability of the current EBM criteria and suggests that the definition of EBM should be reconsidered and refined to ensure accurate and consistent diagnosis across different clinical presentations.

The significance of EBM lies in its association with EBM disease, for which there are perhaps over 30 causes of EBM. These causes are categorized as “dysplasias and dysostoses” like the genetic defect [10] and osteopetrosis [12], “metabolic” such as renal osteodystrophy, and others including degenerative diseases of the spine [4], or hematological disorders. It is not clear whether a Z-score or T-score above + 2.5 or + 4.0 may be a suitable cut-off to find these HBD disorders [13, 14]. In addition, it is unclear whether there is an entirely healthy population with EBM that does not suffer from EBM disorders.

In this study, we refined the criterion that EBM subjects’ T-scores should be ≥ −1.0 to rule out osteoporosis or osteopenia and compared the rate of pathology between subjects with a Z-score ≥ 4.0 and a Z-score between + 2.5 and + 4.0 and found the risk factors for EBM.

Methods

Participant criteria

This retrospective study was approved by the Ethics Committee of Taizhou Hospital (approval number: KL20240507) and was conducted at Taizhou Hospital in Linhai, Zhejiang Province, China, using BMD data collected from 2021 to 2023 at the DEXA center. The inclusion criteria were as follows: Participants aged ≥ 18 years who had undergone at least one DEXA scan to assess their BMD, with complete measurements at the lumbar spine (L1-L4), femoral neck and total hip, and with available data for both Z-scores and T-scores, values meeting the criteria for EBM, as defined below. Exclusion criteria included participants aged < 18 years, participants with repeat BMD measurements, incomplete measurement sites or DEXA results, or participants who had undergone lumbar instrumental fusion, bone cement injection, laminectomy or total hip arthroplasty (THA) prior to DEXA measurement (Fig. 1).

Fig. 1.

Flowchart diagram for selection of participants

DEXA measurement

Two Lunar Prodigy Advance DEXAs (GE Lunar, Madison, WI, USA) were used for areal BMD (aBMD) collection. Weekly cross-calibration was done by two trained technicians from the same department through water film calibration [15]. Routine quality assurance (QA) testing was performed on each machine every morning before the patient was measured. After approximately 100 patients were measured, 30 patients underwent two measurements by the Least Significant Change (LSC) method with re-positioning for precision assessment [16]. Additionally, quality control (QC) tests are conducted weekly to ensure measurement accuracy and precision. At measurement time, all participants were wearing underwear and no metal accessories. The reference criteria of DEXA were based on Chinese population database.

Classification criteria for bone mass

The WHO-NBM was defined as: T-scores of the lumbar spine, femoral neck, and total hip all ≥ −1.0. In our study, NBM was defined as individuals who met the WHO-NBM criteria but were not classified as EBM. The EBM cohort in our study was defined as NBM plus Z-scores ≥ + 2.5 at either the lumbar spine (Z-lumbar), femoral neck (Z-neck) or total hip (Z-hip). Study participants were further divided into two groups: the Moderate EBM (MEBM) group with Z-scores at the lumbar spine, femoral neck or total hip between 2.5 and 4.0 and the Super EBM (SEBM) group with Z-scores ≥ 4.0 at the lumbar spine, femoral neck or total hip (Fig. 2A). Compared to the previous Z-score-based EBM classification, our modified EBM criteria (Z ≥ 2.5 and T ≥ -1.0) yielded a more specific group, excluding individuals with osteoporosis or osteopenia (Fig. 2B).

Fig. 2.

Modified EBM Criteria and Their Relationship with Osteoporosis/Osteopenia. A Illustration of criteria for the osteopenia/osteoporosis and modified criteria for the EBM and its subgroups: Black dashed lines with arrows represent the Z-score scale; the orange solid line represents T-scores. WHO osteoporosis criteria: T ≤ −2.5, WHO osteopenia criteria: −2.5 < T-score < −1.0; WHO-defined NBM: T-scores ≥ −1.0 (yellow area); Our study-defined NBM: T-scores ≥ −1.0 and Z-scores ≤ 2.5 (red area); MEBM: T-scores ≥ −1.0 and 2.5 ≤ Z-scores < 4.0 (gray area); SEBM: T-scores ≥ −1.0 and Z-scores ≥ 4.0 (blue area); Modified EBM: T-scores ≥ −1.0 and Z-scores ≥ 2.5 (gray and blue areas). B Venn Diagram of EBM and Modified EBM Criteria. The figure compares the original EBM criteria (left) with the modified EBM criteria (right). Under the original EBM criteria (Z-scores ≥ 2.5), individuals can be classified as both osteoporosis/osteopenia and EBM, resulting in overlap between the groups. However, with the modified EBM criteria (Z-scores ≥ 2.5 and T-scores ≥ −1.0), there is no overlap between the different groups, allowing for a clearer distinction

Data collection and assessment of comorbidity

Basic demographic characteristics, including age, sex, height, weight and body mass index (BMI), were collected using DEXA data. Medical history, laboratory test results, and radiologic images were collected through the hospital information system (HIS) of Taizhou Hospital and the Taizhou Medical Records Cloud, which is shared by all local hospitals. Imaging modalities such as lung computed tomography (CT) scans, abdominal CT, positron emission tomography-CT (PET-CT), and all spine images were recorded to identify comorbidities such as diffuse idiopathic skeletal hyperostosis (DISH), ankylosing spondylitis (AS), the extent of lumbar degenerative disease (LDD), and malignancies such as hematologic cancers and metastatic prostate, breast, and lung cancers. Laboratory reports were also reviewed to determine the presence of chronic kidney disease (CKD) and type 2 diabetes.

Identification of DISH, AS and LDD

The diagnosis of DISH was made using the criteria of Resnick Donald and Niwayama Gen [17] based on the radiographic appearance, including bridging of four adjacent vertebral bodies by newly formed bone, without severe loss of disk height and without degeneration of the apophyseal and sacroiliac joints [18]. LDD is graded using the Kellgren-Lawrence (KL) classification, a semi-quantitative method. Radiographs are analyzed for lumbar spondylosis based on KL grades as follows: Degree 0, no osteophytes; Degree 1, minimal osteophytes; Degree 2, moderate osteophytes; Degree 3, moderate disc space narrowing with osteophytes; and Degree 4, severe disc space narrowing with large osteophytes and bone sclerosis [19].

Chi-squared Automatic Interaction Detector (CHAID) decision tree mode (DTM)

CHAID DTM was established with SPSS version 26.0 (IBM, New York, American). It was used to identify potential factors which was analyzed by Chi-square test and T-test, and to determine their association with increased BMD in EBM individuals. The dependent variable was the categorization of EBM (MEBM/SEBM), and the risk factors were sex, BMI, LDD severity and the prevalence of DISH and diabetes (age, the prevalence of AS, LDD and history of stroke were excluded).

CHAID can efficiently handle both categorical and continuous variables, automatically identifying important predictors and interactions. This enables it to have predictive capabilities [20]. The result generated a visualized figure resembling “tree”, which represented a decision-making framework. The “branches” of the tree represented different decisions based on the analyzed variables. The classification rules are as follows: (1) Tree growth: the significance level of growth “branch” splitting by α_merge = α_split = 0.05; (2) Tree termination condition: the maximum depth of the tree is 3, the minimum sample number of parent nodes is 50, the minimum sample number of child nodes is 25, and the stopping rule is α = 0.05. If the sample size on the node does not meet the requirements, the node is a terminal node and will not be overflowed.

Statistics

Continuous variables were presented as mean and standard deviation (SD), while categorical variables were presented as number (n) and percentage (%). Differences in age, height, weight, BMI, Z-scores and T-scores were compared using independent samples t-test. Gender composition and proportion of comorbidities were analyzed using the Chi-square test. Variables with significant differences were included in a CHAID analysis to explore predictive factors of EBM. To further investigate possible bivariate associations, two correlation matrixes were constructed based on the Pearson correlation coefficient and the significance of the two-tailed test. All statistical analyses were performed using SPSS version 26.0, with a significance level of p < 0.05. The scatter plots were created using GraphPad prism 9.1.1 software.

Results

Participants

A total of 39,223 DEXA scans were conducted between 2021 and 2023. People under the age of 18 (1,067), repeated ID (4,518), and those who had undergone surgeries affecting BMD (159) were initially excluded. Of the 33,479 eligible participants, those with a lumbar spine, neck or hip T-score below −1.0 (15,864) were excluded (osteoporosis or osteopenia). The remaining 17,615 participants were considered to have WHO-NBM (WHO criteria) whose lumbar spine, neck, and hip T-scores were all ≥ −1.0. In this study, NBM was further defined as individuals who met WHO criteria but had lumbar, neck, and hip Z-scores ≤ 2.5, for a total of 16,143 participants. Finally, 1,472 participants with EBM were identified with a prevalence of 4.4% (1,472/33,479). The EBM group was subdivided into an MEBM of 1,267 with a prevalence of 3.8% and an SEBM of 205 with a prevalence of 0.6% (Fig. 1) and (Fig. 3).

Fig. 3.

Distribution of Different BMD Categories. Corresponding BMD categories use identical colors across (A) and (B). A Pie chart illustrating the percentage distribution of the osteoporosis, osteopenia, NBM, and EBM subgroups (MEBM and SEBM). The denominator is the total number of participants included (n = 33,479). B Icon array graph showing the proportion of individuals in the osteoporosis, osteopenia, NBM, MEBM, and SEBM categories within a sample of 100 people

Comparison of general characteristics between MEBM and SEBM groups

Of the total of 1,472 study participants, 951 were men (65%) with an average age of 59.7 ± 10.5 years. The SEBM group had significantly higher Z-scores and T-scores of all DEXA sites than the MEBM group (p < 0.001) (Supplementary Table 1). The mean age of the SEBM group (62.8 years) is higher than that of the MEBM group (59.4 years) (p = 0.002) (Table 1). In addition, the SEBM group had a higher proportion of men (76% vs. 60% in the MEBM) and a higher BMI than the MEBM groups (p < 0.001). With the increase of age, the proportion of men (75% vs. 51% vs. 41%, p < 0.001) and the average BMI (25.7 vs. 25.2 vs. 23.9, p < 0.001) is also increasing (Supplement Table 2).

Table 1.

Comparison of the characteristics between MEBM and SEBM groups

	Total (n = 1472)	MEBM (n = 1267)	SEBM (n = 205)	p-value
Age – years	59.7 ± 10.5	59.4 ± 10.3	62.8 ± 11.4	0.002
Men – n (%)	951 (65)	756 (60)	156 (76)	< 0.001*
BMI	25.4 ± 3.2	25.3 ± 3.2	26.5 ± 3.3	< 0.001
Underweight – n (%)	15 (1.0)	13 (1.0)	2 (1.0)	< 0.001*
Normal weight – n (%)	751 (51)	628 (50)	71 (35)
Overweight – n (%)	665 (45)	538 (42)	97 (47)
Obesity – n (%)	128 (8.7)	88 (7.0)	35 (17)
Comorbidities – n (%)
DISH – n (%)	290 (20)	215 (17)	75 (37)	< 0.001
AS – n (%)	10 (0.7)	4 (0.3)	6 (2.9)	< 0.001
Prevalence of LDD – n (%)	897 (61)	740 (58)	157 (77)	< 0.001
LDD severity – n (%)				< 0.001
Degree 1	81 (5.5)	70 (5.5)	11 (5.4)
Degree 2	212 (14)	172 (14)	40 (20)
Degree 3	358 (24)	294 (23)	64 (31)
Degree 4	246 (17)	204 (16)	42 (20)
Malignancy – n (%)				0.13
Hematology	8 (0.5)	5 (0.4)	3 (1.5)
Prostate cancer	5 (0.3)	5 (0.4)	0 (0)
Breast cancer	15 (1.0)	13 (1.0)	2 (1.0)
Lung cancer	29 (2.0)	25 (2.0)	4 (2.0)
Other	51 (3.5)	39 (3.1)	12 (7.3)
CKD – n (%)	133 (9.0)	106 (8.4)	27 (13)	0.07
Stroke – n (%)	102 (6.9)	80 (6.3)	22 (11)	0.02
Diabetes – n (%)	251 (17)	197 (16)	54 (26)	< 0.001
Thyroid dysfunction – n (%)	47 (3.2)	40 (3.2)	7 (3.4)	0.85
Parathyroid dysfunction – n (%)	27 (1.8)	24 (1.9)	3 (1.5)	0.67
Rheumatoid arthritis – n (%)	14 (1.0)	12 (1.0)	2 (0.9)	0.97
Other immune related diseases – n (%)	18 (1.2)	15 (1.2)	3 (1.5)	0.74

P-values were determined using the Pearson Chi-squared test^*, and independent-samples T test for the others (MEBM vs. SEBM). Total = EBM. Underweight: BMI < 18.5; Normal weight: 18.5 ≤ BMI < 25; Overweight: 25 ≤ BMI < 30; Obesity: BMI ≥ 30. Abbreviations: EBM = Elevated Bone Mass; MEBM = Moderate Elevated Bone Mass; SEBM = Super Elevated Bone Mass; BMI = Body Mass Index; DISH = Diffuse Idiopathic Skeletal Hyperostosis; AS = Ankylosing Spondylitis; LDD = Lumbar Degenerative Disease; CKD = Chronic Kidney Disease

Comparison of the clinical characteristics between MEBM and SEBM groups

SEBM group had higher prevalence of DISH (37% vs. 17%), AS (2.9% vs. 0.3%), LDD (77% vs. 58%), and diabetes (26% vs. 16%) (all p < 0.001) compared to MEBM group. LDD severity was also more pronounced in the SEBM group, with more participants in higher severity degrees (degree 3 and 4). The proportion of individuals with a history of stroke was higher in SEBM (11% vs. 6.3%, p = 0.02). There were no significant differences in the overall prevalence of malignancies, CKD, thyroid or parathyroid dysfunction, rheumatoid arthritis, or other immune-related diseases between the groups (Table 1). In addition, we categorized individuals with EBM into elderly, middle-aged, and young groups, and found that the elderly group had a significantly higher prevalence of DISH (33% vs. 8.2% vs. 0%), LDD (80% vs. 46% vs. 21%), stroke (13% vs. 1.7% vs. 0%), and diabetes (24% vs. 11% vs. 6.9%) compared to the middle-aged and young groups (all p < 0.001) (Supplement Table 2).

Different correlation trends in MEBM and SEBM groups

Individuals with a lower BMI generally have a higher risk of low BMD. As expected, neck T-score, hip T-score, and L1-4 T-score were positively correlated with BMI in the MEBM group (r = 0.12, r = 0.25, r = 0.29, respectively; all p < 0.001). However, no significant correlation was observed in the SEBM group (p > 0.05). Conversely, in the SEBM group, L1-4 Z-score and neck Z-score showed a weak negative correlation with BMI (r = −0.15, p = 0.03; r = −0.17, p = 0.02), while no significant correlation of Z-scores with BMI was observed in the MEBM group (p > 0.05) (Fig. 4A).

Fig. 4.

Correlations between BMD and related factors in EBM individuals. Correlation coefficients and sig. two-sided (indicated by *: p < 0.05 and **: p < 0.01) are marked on each part. A The heatmap shows continuous variables on both axes, with each square corresponding to combinations of horizontal and vertical variables. Diagonal squares show the correlation of a variable with itself (correlation coefficient is + 1), while off-diagonal ones show correlations between different variables. Each number represents the relevant correlation coefficient, and color intensity indicates the strength of correlation, with deeper red for stronger positive correlation and deeper blue for stronger negative correlation. Additionally, differences between MEBM and SEBM are statistically significant (*: p < 0.05; **: p < 0.001). B Scatter plot of T score (L1-4) vs. T score (Neck) in MEBM (r = 0.24, p < 0.001) and SEBM groups (r = −0.14, p = 0.04), showing significantly different correlation coefficients between groups. C Scatter plot of T score (L1-4) vs. T score (Hip) in MEBM (r = 0.30, p < 0.001) and SEBM groups (r = −0.02, p = 0.78), also showing significantly different correlation coefficients between groups

BMD generally decreases with age. As expected, in the MEBM group, L1-4 T-score was moderately negatively correlated with age (r = −0.29, p < 0.001). However, in the SEBM group, no significant correlation between L1-4 T-score and age was observed (p > 0.05) (Fig. 4A).

If BMD is generally high or low at all sites, the T-score or Z-score at one site should be positively associated with values at other sites. As expected, in the MEBM group, neck T-score and hip T-score were positively associated with L1-4 T-score (r = 0.24, r = 0.30, respectively; all p < 0.001) (Fig. 4C). In contrast, in the SEBM group, a negative correlation was observed between neck T-score and L1-4 T-score (r = −0.14; p = 0.04) (Fig. 4B).

DTM of the Characteristics Associated with the Subgroup analysis in EBM individuals

The final decision tree (Fig. 5) had 4 levels and 10 nodes, 6 were terminal nodes, and 4 splits. From the 9 clinical characters initially included in the modelling process, 5 were automatically selected by the algorithm and included in the final tree: the risk factors were sex, BMI, LDD severity and the prevalence of DISH and diabetes. The DTM analysis using CHAID identified the prevalence of DISH as the most important predictor for SEBM, meaning it is the primary factor influencing whether an individual belongs to the SEBM group. And followed by the LDD severity, BMI and the prevalence of diabetes.

Fig. 5.

Using CHAID DTM identify the factors that affect SEBM event. Level 1: Node 0 represents the total group. Level 2: “Yes” indicates EBM with DISH, and “no” indicates EBM without DISH, generating node 1 (EBM with DISH) and node 2 (EBM without DISH). Level 3: Based on node 1, node 3 (EBM with DISH and LDD of degree 3—4) and node 4 (EBM with DISH and LDD of degree 0—2) are categorized by the severity of LDD. Categorized by BMI, node 5 (EBM without DISH, BMI ≤ 24.4), node 6 (EBM without DISH, 24.4 < BMI < 29.5), and node 7 (EBM without DISH, BMI ≥ 29.5) are derived from node 2. Level 4: Based on node 6, node 8 (EBM without DISH, 24.4 < BMI < 29.5, with diabetes) and node 9 (EBM without DISH, 24.4 < BMI < 29.5, without diabetes) are determined

When the analysis split the population (Node 0) into two groups according to the prevalence of DISH: those with DISH (Node 1) exhibited a higher rate of SEBM (26%) compared to those without DISH (Node 2) (11%). Node 1 was split into two child nodes (Nodes 3–4) according to the severity of LDD. When LDD severity of degree 3–4 (Node 3), the probability of SEBM was greater (39%) than that with LDD severity of degree 0–2 (Node 4) (19%). And then Node 2 was split into three child nodes (Nodes 5–7) according to the BMI. When BMI ≥ 29.5 (Node 7), the probability of SEBM was greater (22%) than that with 24.4 < BMI < 29.5 (Node 6) (13%) and BMI ≤ 24.4 (Node 5) (7.1%). Finally, Node 6 was split into two nodes (Nodes 8–9) according to the prevalence of diabetes. When individuals with diabetes (Node 8), the probability of SEBM was greater (19%) than that with non-diabetes (Node 9) (11%).

Severe LDD (degree 3–4) significantly influences SEBM rate in DISH patients. Additionally, a BMI ≥ 29.5 is influential for SEBM in non-DISH patients, and diabetes is a risk for SEBM in non-DISH patients with 24.4 < BMI < 29.5 (Fig. 5).

Discussion

This study found a prevalence of EBM of 4.3% with a Z-score cutoff of ≥ 2.5 and 0.6% with a cutoff of Z ≥ 4.0. In comparison, two studies from France that used a Z-score cutoff of ≥ 4.0 reported EBM prevalence of 1.26% and 1.43% [4, 6]. A Canadian study defined EBM in women aged ≥ 50 years using T-score ≥ 2.5 or Z-score ≥ 2.0, but the criteria identified different populations, with hip Z-score ≥ 2.0 identifying 5.6% and hip T-score ≥ 2.5 only 0.3% [7]. The different prevalence of EBM with a Z-score ≥ 2.5 or Z-score ≥ 4.0 across clinical trials may be attributed to several factors, including variations in the gender and age composition of the participants, ethical differences, the use of multiple Z-scores instead of a single Z-score, or the application of different classification criteria.

Our study found that some participants who met the previous EBM criteria may also qualify for a diagnosis of osteoporosis or osteopenia. Previous studies established thresholds of T-scores or Z-scores ≥  + 4.0 or + 2.5 at the lumbar spine, neck, or hip to detect EBM [4, 6–8]. However, these criteria did not exclude individuals with T- or Z-scores below −1.0 at any site such as the lumbar spine, femoral neck, or total hip — individuals who could be diagnosed with osteopenia or osteoporosis according to current WHO standards [4, 6–8]. Our study found that individuals who fulfill the EBM criteria may also be diagnosed with osteoporosis or osteopenia according to the WHO criteria (Fig. 6). To counteract this, our EBM criteria explicitly exclude individuals with high Z-scores who have also been diagnosed with osteoporosis or osteopenia according to the WHO criteria. Most clinicians who read DEXA reports are doctors treating osteoporosis and primarily rely on T-scores, often overlooking high Z-scores, especially in men over 50 years old and women in menopause. Our modified EBM criteria will encourage clinicians to consider both T-scores and Z-scores on DEXA reports, thereby bridging the research areas of clinicians focused on either osteoporosis or EBM.

Fig. 6.

Case Report: DEXA of an 82-year-old woman meeting both WHO T-score criteria for osteoporosis and EBM criteria based on Z-score

In our study, the average age of EBM subjects is 59.7 years and the age of SEBM is higher than that of MEBM. Contrary to the general assumption that EBM individuals are typically young, our data are consistent with previous studies showing that people with EBM are predominantly older rather than young adults. A study from France found that 55% of EBM individuals were aged 60 years or older [4]. A UK study also reported the average age of EBM to be 64.5 years [11]. For example, these causes account for the majority of EBM cases (80%), with LDD being the most common at 65% of older people [4]. Second, in agreement with previous studies [4, 6, 11], we used Z-scores to classify EBM. The Z-score compares a person's bone density with the expected average for their age, sex and height, while the T-score compares bone density with that of a young, healthy adult of the same sex. The aim was to find out why people with EBM have a higher bone mass than their peers. Ultimately, the majority of individuals who underwent DEXA measurements were generally older, and only a small proportion of younger individuals had DEXA measurements.

Our study showed that people in the SEBM group had a higher proportion of men and a higher BMI (with more people being overweight or obese) than those in the MEBM group. In contrast, studies from France showed that women predominated in the EBM population, at 64% or 56% depending on the study [4, 6]. A British study reported that 78.9% of people with EBM were women [11]. One possible explanation for this is that more women than men undergo DEXA screening for osteoporosis. Secondly, the selection criteria for EBM were different among studies. Thirdly, ethnic differences could also contribute, although no specific ethnic variations have been identified in previous studies [4, 7]. Finally, in our study, the percentage of obese individuals was low at 8.7%, whereas in the EBM population in France the obesity rate ranged from 41 to 44% [4, 6] and in the UK it was 31.0% [11].

The SEBM group showed abnormal patterns, as Z-scores showed a negative correlation with BMI and no correlation with age. In addition, the L1-4 T-score was negatively correlated with the neck T-score. Numerous clinical trials have demonstrated the association between bone loss and advancing age [21, 22]. A cross-sectional study using the NHANES concluded that BMD is positively correlated with BMI [23], which may be attributed to the greater mechanical stress of body weight influencing bone formation through biomechanical overload [24, 25]. The abnormal patterns observed in the SEBM group may be attributed to the increased proportion of DISH, which typically affects lumbar T or Z values. Previous studies have confirmed that factors such as DISH and LDD, which contribute to osteophytic hyperplasia, may lead to increased BMD [26–28]. In conclusion, the high prevalence of DISH in the SEBM group likely affected BMD measurements.

Our DTM model identified three factors that may lead to SEBM: DISH with severe LDD, non-DISH with a BMI ≥ 29.5, and diabetic non-DISH patients with a BMI between 24.4 and 29.5. Previous studies have suggested that obesity may act as a mediator and influence both diabetes and increased BMD [29]. DISH is typically associated with older age, male gender, obesity, hypertension, and diabetes mellitus [30]. Previous research has shown that obese people are more likely to develop DISH at a young age, before the age of 50 [31]. DISH primarily affects the lower thoracic spine, but can also extend to the lumbar spine and increase the risk of LDD [32]. Our findings highlight the complex interplay between metabolic disorders and skeletal health and suggest that identifying SEBM may improve our understanding of these risks and guide our interventions.

In our study, we use the term EBM to categorize the population with high Z-scores on DEXA reports. The term HBM is usually linked to genetic mutations, especially in genes like SOST, LRP5 or LRP6 [10, 33, 34]. Giammarco De Mattia's case report on LRP5 high bone mass highlights the distinction between HBM as a category of disorders and EBM as a phenomenon detected through imaging techniques like DEXA [35]. Additionally, two studies in France specifically identified individuals with EBM by DEXA and classified them as EBM [4, 6]. Consequently, while EBM is associated with increased BMD on DEXA, HBM specifically refers to conditions, such as those resulting from genetic mutations.

This study provides several clinical recommendations. First, we recommend that DEXA reports incorporate specific criteria for EBM, rather than simply labeling BMD as normal, to prevent important underlying conditions from being missed. Second, we support clinicians in reconsidering the WHO criteria for BMD ≥ −1.0 as normal in clinical practice, as a BMD T-score ≥ −1.0 is not necessarily normal. We suggest that individuals with a BMD T-score ≥ −1.0 and a Z-score < 2.5 should be classified as NBM, as this population does not fall under the categories of osteoporosis, osteopenia, or EBM. Third, EBM should serve as a red flag for clinicians, as it often indicates potential underlying conditions such as artefactual causes [36] (e.g., factors affecting DEXA measurements such as DISH, AS, or LDD), malignancy, genetic diseases, chronic kidney disease, and many other conditions. However, we believe that EBM presents a much more intriguing and counterintuitive research trend, especially in the context of osteoporosis or osteomalacia, as well as genetic osteomalacia. Additionally, the most frequent causes of EBM, such as DISH and LDD, varied across older, middle-aged, and younger groups in our study. This suggests that future EBM classification criteria should be more specific and adapted based on age. Finally, DEXA is not a perfect tool for diagnosing EBM, as artefactual causes may affect the true BMD. For example, our DTM model identified three factors that may contribute to SEBM: DISH with severe LDD, non-DISH with a BMI ≥ 29.5, and diabetic non-DISH patients with a BMI between 24.4 and 29.5.

Although DEXA-measured BMD is the gold standard for diagnosing osteoporosis or osteopenia [37], there are several limitations in using DEXA BMD alone for defining EBM. First, DEXA typically measures the lumbar spine and proximal femur, but marrow disorders, often hematologic in nature, can cause axial osteosclerosis [38]. These conditions may not be detected in the proximal femur, especially in adults with fat in the femoral marrow [39]. The symmetric skeletal dysplasias and dysostoses can lead to asymmetric BMD elevations on DEXA [40]. It is essential to recognize that DEXA measures aBMD, not volumetric BMD (vBMD), and that the results can be influenced by bone geometry [41]. Furthermore, DEXA results should always be interpreted in the context of the patient’s clinical characteristics and medical history. In healthy populations with EBM, individuals have been identified as carriers of certain genetic mutations, such as van Buchem disease, sclerosteosis, or mild mutations in LRP5 and LRP6 [13]. Future studies should adopt a dual-assessment approach that integrates various imaging modalities, as this could improve our understanding of BMD in EBM and help establish more standardized definitions of EBM across different imaging techniques.

Our research has several limitations. The DEXA data were collected in a single hospital, which limits generalizability. The EBM group was not compared with the NBM, osteopenia, and osteoporosis groups. In addition, the determination of the severity of DISH and LDD was subjective, potentially introducing bias. Additionally, our study lacked sufficient data on confounding factors about EBM causes, such as genetic diseases.

Conclusions

Using the modified criteria for EBM, we exclude cases of osteoporosis and osteopenia from EBM. Our study identified key factors contributing to SEBM. The DTM model highlighted DISH with severe LDD, non-DISH with high BMI, and diabetes as important risk factors contributing to SEBM. These findings emphasize the importance of considering underlying conditions when evaluating EMB.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

The study was supported by the Natural Science Foundation of Zhejiang Province (Grant Number: TGY24H250007) and the National Natural Science Foundation of China (Grant Number: 82202738).

Data Availability

The authors are willing to share data related to the research upon request.

Declarations

Ethical standards disclosure

All procedures involving human participants were conducted in accordance with the ethical standards of the institutional and/or national research committees, as well as the 1964 Helsinki Declaration and its later amendments. Ethical approval for the study was obtained from the Ethics Committee of Taizhou Hospital, Zhejiang Province (Approval Number: KL20240507).

Conflict of interest

Jiangjie Chen, Lingling Cao, Chenghao Xu, Fangying Lu, Liwei Zhang, Anpeng Xu, Yahong Chen, Tao-Hsin Tung, and Dun Hong declare that they have no conflict of interest.