Introduction

Osteoporosis is a highly prevalent disease among older adults. As age increases, osteoporosis and the increased risk of falls can lead to fractures, which can severely and negatively affect people’s quality of life and significantly increase the risks of hospitalization and death [1]. The incidence of diabetes is increasing as people’s dietary patterns are shifting toward higher energy levels. Hyperglycemia increases the production of advanced glycation end products and negatively affects bone mineralization, bone remodeling, and bone strength [2]. Diabetic complications can also considerably increase the risk of osteoporosis [3]. The prevalence of diabetic osteoporosis accounts for approximately > 50% of diabetic patients [4]. Moreover, the adverse outcomes of fracture are more severe in diabetic patients than in normoglycemic patients, and therefore, early identification of high-risk groups among elderly T2DM patients is critical [5].

Age, gender, vitamin D, muscle strength, and nutritional status are risk factors for osteoporosis. Elderly people are prone to malnutrition because of their specific metabolic characteristics and disease. Considerable evidence has proven that malnutrition is an independent risk factor for elderly patients with osteoporosis; studies have reported that low body weight, reduced ABL, and prealbumin can lead to an increased incidence of osteoporotic fractures [6, 7]. The geriatric nutritional risk index (GNRI) is used to assess the nutritional condition of older adults. It is calculated on the basis of serum ABL levels and the ratio of current actual body weight to ideal body weight [8].The GNRI allows for the early detection and diagnosis of malnutrition, timely and appropriate administration of interventions, as well as identification of conditions at risk for adverse effects, including cancer prognosis, postoperative complications, and mortality in patients on dialysis and those with cardiovascular disease [9,10,11,12,13]. Moreover, this tool is highly accurate and can be easily used clinically.

To our knowledge, few studies have evaluated the correlation between GNRI and osteoporosis. Bijuan Qing et al. [14] reported that the GNRI is associated with osteoporosis in geriatric patients. Liang Wang et al. [15] revealed that the GNRI is associated with osteoporosis as well as BMD in T2DM patients. Their study population comprised patients from southern China who had adequate vitamin D levels. However, in northern China, the majority of the population has vitamin D insufficiency or deficiency, is of an older age, and has a longer duration of diabetes; therefore, their study is not completely representative of the population with T2DM in China. Hence, the present study investigated the association between the GNRI and the presence of osteoporosis and assessed the value of GNRI for predicting osteoporosis in patients with T2DM.

Materials and methods

Methods

This was a cross-sectional observational study on 610 patients with T2DM of age ≥ 60 years who were treated at the Hebei General Hospital from January 2018 to December 2020. The diagnostic criteria of T2DM were based on the 1999 WHO. The study exclusion criteria for the participants were as follows: (1) individuals with diseases affecting the bone metabolism or the nutritional status, such as malignancies, severe liver diseases, kidney diseases, pituitary-related diseases, thyroid and parathyroid diseases, adrenal diseases, rheumatoid arthritis, and acute inflammatory diseases; (2) individuals who have been bedridden for a long period; (3) individuals who are taking drugs that affect their bone metabolisms, such as vitamin D, calcium, bisphosphonates, and glucocorticoids. This study protocol was approved by the Ethics Committee of the Hebei General Hospital and complies with the Declaration of Helsinki.

Clinical information

Patient demographics and clinical characteristics, including information on gender, age, disease duration, and comorbidities, were collected from their respective medical records.

The weights of the patients were measured while they were wearing light clothing and their heights were measured without shoes. The body weight obtained was divided by height squared (kg/m2) to calculate the body mass index for each patient.

Biochemical indicators

Serum samples were collected after fasting for at least 8 h. Triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and albumin levels, fasting blood glucose, glycosylated hemoglobin, bone metabolism indicators, ALP, bone glaprotein (BGP), β-CTX, P1NP, 25(OH)D, and PTH were measured. In addition, other biochemical markers, such as uric acid, blood creatinine, and calcium were also analyzed.

Bone mineral density

Bone densitometry was performed using a dual-energy X-ray bone densitometer to measure the bone density values in the lumbar spine (L1–4), femoral neck, and total hip. In accordance with the criteria for the definition of osteoporosis in 1994 WHO, T values ≤ -2.5 standard deviations were obtained for any part of the lumbar spine, femoral neck, or total hip [16].

Calculating GNRI

The GNRI was calculated using the following formula:

  • GNRI = [1.489 × albumin (g/dL)] + [41.7 × (body weight/WL0)].

WL0 represented the ideal body weight (kg), which was calculated as follows:

  • For men: height (cm)-100-[(height (cm)-150)/4].

  • For women: height (cm)-100-[(height (cm)-150)/2.5].

  • Actual weight divided by the ideal weight was set to 1 where the actual weight exceeded the ideal weight.

According to the results obtained, there are 4 levels: GNRI < 82 indicating a high nutritional risk, GNRI:82 to < 92 indicating moderate nutritional risk, GNRI:92 to ≤ 98 indicating low risk, and GNRI > 98 indicating no risk.

Statistical analysis

All data were statistically analyzed using the SPSS 25 software. Data distribution was evaluated with the Kolmogorov–Smirnov test. The mean ± standard deviation was applied to indicate that the data were subject to a normal distribution, while analysis of variance (ANOVA) was applied for comparisons between groups. The median (25th percentile, 75th percentile) was applied to indicate that the data did not conform to a normal distribution, and Mann–Whitney U-test was applied for comparisons between the groups. Categorical data were expressed as frequencies (%), and differences between the groups were determined using the χ2 test. Spearman correlational analysis was applied to determine the correlation among GNRI, BMD, and bone metabolism indicators. Logistic regression analysis was performed to assess the relationship between GNRI and osteoporosis. ROC curves were applied to assess the predictive properties of GNRI for osteoporosis and to calculate the area under the receiver operating characteristic curve (AUC).

Results

Clinical characteristics of the enrolled patients

As shown in Fig. 1. Selection of Subjects，all patients were grouped according to the GNRI score and categorized into the no-nutritional risk group (GNRI > 98) and the nutritional risk group (GNRI ≤ 98) [17].

Fig. 1
figure 1

Flow chart depicting the subjects’ selection process

Full size image

As shown in Table 1, when compared with the no-nutritional risk group, the nutritional risk group had a longer diabetes course, older age, lower BMI, lower ABL, higher HbA1c, lower TG, lower Ca, lower 25 (OH) D, lower PTH, lower femoral neck BMD, lower total hip BMD and higher prevalence of osteoporosis. The differences between the two groups were statistically significant (P < 0.05).

Table 1 Clinical characteristics of patients as stratified by their GNRI scores
Full size table

The patients were grouped according to the presence and absence of osteoporosis. Patients in the non-osteoporotic group had higher GNRI values when compared to those in the osteoporotic group (P < 0.05; Fig. 2).

Fig. 2
figure 2

The chart depicting the GNRI score of the non-osteoporotic and osteoporotic groups

Full size image

Spearman's correlations between the geriatric nutritional risk index and bone metabolism indicators

Spearman's correlation analysis yielded a positive correlation between GNRI and Ca, 25(OH)D, PTH. A negative correlation was recorded between GNRI and PINP. After adjusting for confounding factors, including age, the duration of diabetes, HbA1c, TC, TG, HDL, LDL, UA, Cr, and 24 h microalbuminuria (24 h-mAlb), GNRI was positively correlated with Ca, 25(OH)D, and PTH and negatively correlated with ALP and PINP (Table 2).

Table 2 Correlation between the Geriatric Nutrition Risk Index and the indicators of bone metabolism
Full size table

Correlation analysis revealed that GNRI was positively correlated with lumbar BMD, femoral neck BMD, and total hip BMD (P < 0.05; Fig. 3).

Fig. 3
figure 3

Scatter diagrams depicting the correlation between GNRI and BMD

Full size image

Logistic regression analysis of participants with osteoporosis

The results of the univariate logistic analysis suggest that GNRI, gender, age, duration of diabetes, 24 h-mAlb, 25(OH)D, P1NP, and FPG were associated with osteoporosis (Table 3).

Table 3 Univariate Logistic regression analysis of osteoporosis
Full size table

Multivariate analysis of the association between GNRI and osteoporosis through logistic regression (Table 4). After adjusting for gender, age, the duration of diabetes, FPG, 25(OH)D, P1NP, and 24 h-mAlb a significant association was established between GNRI and osteoporosis.

Table 4 Multivariate logistic regression analysis of osteoporosis
Full size table

Using GNRI as the categorical variable, the risk of osteoporosis in the nutritional risk group was 3.31-times higher than that in the non-nutritional risk group (Table 5).

Table 5 Multivariate logistic regression analysis of osteoporosis
Full size table

Predictive properties of GNRI for osteoporosis

ROC curve analysis was performed with GNRI as the test variable and the presence of osteoporosis as the status variable (Fig. 4). The analysis yielded an area under the curve for GNRI of 0.695, 95% confidence intervals of (0.647,0.743), with an optimal GNRI threshold of 99.56 for predicting osteoporosis, and a sensitivity of 81.19% and specificity of 52.53%.

Fig. 4
figure 4

Receiver operating characteristic curve of osteoporosis

Full size image

Discussion

The GNRI is an indicator of nutritional status in elderly people. It is calculated using serum ABL level, weight, and height. Moreover, it involves a dual assessment of serum ABL and BMI that complements and improves its diagnostic accuracy. Good nutritional status plays a good role in bone metabolism. Similarly, malnutrition increases the risk of osteoporosis and fragility fractures [8].

Therefore, awareness is required to prevent complications associated with malnutrition in elderly people with high fracture risk. Several recent studies on GNRI and osteoporosis have reported a significant correlation between the two. Tokumoto et al. reported that a lower GNRI value is a risk factor for femoral neck BMD in patients receiving modified biological disease antirheumatic drugs for rheumatoid arthritis (RA) [18]. A positive correlation was observed between the GNRI and BMD as well as the T-score in patients undergoing hemodialysis and total thyroidectomy [19, 20]. In a study[21], 57% of 858 women with with hip fractures were at a high or moderate risk of nutrition. Overall, the GNRI has the potential to early identify patients at a high risk of osteoporosis among the elderly T2DM population.

In the present study, we investigated the correlation between GNRI and osteoporosis in T2DM patients from northern China. Our results yield that low levels of GNRI have a negative effect on bone metabolism in patients. According to the correlation analysis, the GNRI was positively correlated with BMD (P < 0.05). The logistic regression analysis revealed a significant association of the GNRI with osteoporosis. The ROC analysis yielded a GNRI cut-off value of 99.56. Our results providing a theoretical basis for screening for osteoporosis clinically.

The current understanding of the possible mechanism underlying the association between the GNRI and osteoporosis is as follows. First, malnutrition affects calcium and vitamin D intake, which may increase bone mineral loss in patients, making bone mineralization difficult and leading to osteoporosis. Second, hypoalbuminemia is a marker of both nutritional status and chronic inflammatory response. Hypoalbuminemia activates osteoclasts and inhibits osteoblasts through NF-κB factors and other inflammatory cytokines[22]. Moreover, hypoalbuminemia causes a decrease in insulin-like growth factor-1 synthesis, thereby leading to a decreased number of osteoblasts, decreased cellular activity, increased osteoclast lifespan, increased bone resorption, and decreased bone remodeling [23]. Finally, hypoproteinemia is associated with inadequate muscle synthesis and decreased skeletal muscle mass, resulting in decreased balance and gait capacity, which is associated with the risk of falls and fractures [24, 25].

Liang Wang et al. found no correlation between GNRI and 25(OH)D. By contrast, we yielded a positive correlation between GNRI and 25(OH)D. A possible reason for our different results is that in southern China even though nutritional status indicates reduced dietary vitamin intake, vitamin D levels can still be ensured with adequate sun exposure. By contrast, in northern China, insufficient sun exposure and nutritional barriers make people more prone to vitamin D insufficiency and deficiency. Studies have shown a close relationship between latitude sunlight deficiency, skin coverage, and vitamin D. In China, a clear geographical division of vitamin D deficiency can be seen, with populations in northern, northeastern, and northwestern China at the north of 35 degrees north latitude being more severely undernourished, whereas vitamin D levels are adequate in areas south of 25 degrees north latitude and the middle of the country [26, 27].

In our study, GNRI was negatively correlated with P1NP. A possible reason for this negative correlation is that PINP secreted by osteoblasts is a known marker of bone formation and reflects collagen formation and osteoblast activation. During bone turnover, bone formation and resorption are tightly coupled, with accelerated bone turnover predisposing a person to bone loss. Therefore, high GNRI values are associated with low bone turnover and reduced bone loss. GNRI was positiveiy correlated with PTH. A possible reason is that a positive correlation was observed between serum PTH and BMI and fat mass. Mehrotra indicated that reduced PTH is a risk factor for malnutrition [28, 29]. PTH promotes the inward flow of calcium ions into adipocytes and stimulates adipose synthesis. Accordingly, low PTH levels inhibit the adipose synthesis and cause protein depletion. Bone-specific alkaline phosphatase (BALP) is closely related to normal bone growth and development. This enzyme is a marker of osteoblast maturation and activation. However, the specificity of our current assay is not good. Moreover, BALP has some crossover with liver-derived ALP, and we measured total ALP and not BALP. Therefore, the relationship of the GNRI with ALP does not accurately reflect the bone metabolism level.

Logistic regression analysis revealed that advanced age is an independent risk factor for osteoporosis, with bone density decreasing every year with an increase in age. Furthermore, oxidative stress, which increases osteoclast activity and bone resorption, is a cause of age-related bone loss. According to the present study, diabetes duration is an influential factor for osteoporosis in T2DM patients. The relative risk of diabetes course increases from 1.40 (1. 08–1. 82) at less than 5 years to 2.66 (2. 04–3. 47) at greater than 15 years [30]. The non-enzymatic glycosylation response of T2DM contributes to the decline in bone mass. Because of hyperglycemia, advanced glycosylation end products are accumulated in the organic bone matrix, thereby resulting in stiffening of type I collagen in the bone matrix, decreased bone strength, increased bone fragility, and promotion of osteoblast apoptosis [31]. Our results suggest that suboptimal glycemic control is also an independent risk factor for osteoporosis development.

Furthermore, based on our regression analysis of the GNRI and osteoporosis, we concluded that uric acid (UA) is not associated with osteoporosis and is neither a protective nor a risk factor for osteoporosis. Several studies have reported that higher UA levels are protective against osteoporosis [32,33,34]. Our results were different from those of these studies may be because of differences in gender, region, ethnicity, study methodology, and sample size. Finally, the association of serum UA with osteoporosis may be directly or indirectly confounded by the fact that many older adults have two or more chronic diseases, such as obesity and DM.

This study has some limitations. First, this cross-sectional study does not offer a mechanism-related explanation for the observed association, This study also does not indicate a causal relationship between the GNRI and osteoporosis. Second, the serum data of the T2DM patients and BMD were both collected only once, thus leading to bias. Third, some relevant parameters affecting the study results may have been overlooked, such as a history of smoking and alcohol consumption, hormone levels, dietary habits, exercise situation, and history of previous fractures.

In summary, our results demonstrate that a lower GNRI is associated with a higher prevalence of osteoporosis and that the GNRI is an easy-to-use tool for assessing nutritional status and osteoporosis in T2DM patients. Nutritional supplementation therapy may reduce osteoporosis prevalence in T2DM patients.

Conclusions

A lower GNRI among T2DM patients in northern China is associated with a higher prevalence of osteoporosis.