Abstract
Objective: To investigate the association between the basic and clinical characteristics of patients with type 2 diabetes mellitus (T2DM) and their susceptibility to Klebsiella pneumoniae colonization (KPC). Additionally, a clinical prediction model was developed to identify high-risk patients for KPC. Methods: Data from 486 T2DM patients who visited Shanghai Fifth People’s Hospital from December 2020 to December 2022 were retrospectively collected. Patients were classified into the KPC group and normal group based on their Klebsiella pneumoniae test results. Differences between the two groups were analyzed using t-test and chi-square test. Logistic regression was performed to identify factors influencing KPC susceptibility in T2DM patients, with odds ratios (ORs) calculated. A clinical prediction model was constructed using a nomogram and evaluated through the area under the receiver operating characteristic (ROC) curve (AUC), Hosmer-Lemeshow test, calibration curve, and decision curve analysis (DCA). Results: Of the 486 T2DM patients, 124 were found to have KPC, with a colonization rate of 25.51%. Logistic regression analysis revealed that hospitalization within the past six months, elevated white blood cell count, decreased hemoglobin, and elevated ferritin levels were independent risk factors for KPC. Thyroid and liver function indicators were also associated with KPC susceptibility. The clinical prediction model achieved an AUC of 0.74 (95% CI: 0.68-0.80). The calibration curve indicated no significant differences between observed and predicted values, suggesting that the model effectively identifies high-risk KPC patients. Conclusion: T2DM patients are at an increased risk of secondary KPC. Identifying key risk factors for KPC in T2DM patients has significant clinical implications for early identification, targeted interventions, and individualized treatment strategies.
Keywords: Type 2 diabetes mellitus, Klebsiella pneumoniae, colonization, clinical prediction model
Introduction
Type 2 diabetes mellitus (T2DM) is a chronic metabolic disorder characterized by insulin resistance and impaired insulin secretion, leading to persistent hyperglycemia [1]. With rapid global economic development and lifestyle changes, T2DM has become a critical public health challenge, with increasing prevalence and associated mortality [2]. Representing more than 90% of all diabetes cases, T2DM significantly impacts patients’ quality of life and life expectancy due to its severe complications [3]. Studies have established T2DM as a major risk factor for cardiovascular diseases, including heart disease and stroke [4]. Furthermore, prolonged hyperglycemia contributes to chronic kidney disease, retinopathy, neuropathy, and other debilitating complications [5].
Klebsiella pneumoniae (KP), a gram-negative bacterium, is widely present in both community and healthcare settings and is known for its high pathogenicity and multidrug resistance [6]. Recent research highlights diabetes as a key risk factor for KP infections [7]. This underscores the complex interplay between T2DM and infection risk, emphasizing the need for effective prevention and intervention strategies.
Chronic hyperglycemia in T2DM patients often leads to complications such as immune dysfunction, microvascular damage, and neuropathy. These pathological changes increase susceptibility to infections by opportunistic pathogens like KP [8]. The altered microenvironment in diabetic patients, characterized by acidosis and nutrient-rich tissue fluids, facilitates KP colonization and proliferation.
In some T2DM patients, KP can be detected in urine, stool, or respiratory tract samples without manifesting clinical symptoms, a condition termed KPC. Secondary KPC is associated with higher risks of infection and worse clinical outcomes. KPC in T2DM patients not only leads to localized infections, such as pneumonia and urinary tract infections, but may also disseminate systemically through the bloodstream, resulting in severe complications such as sepsis and endophthalmitis [9].
Early identification and management of KPC are crucial for preventing diabetes-related infections and reducing associated morbidity. However, clinicians currently lack effective predictive tools for assessing the risk of KP colonization in T2DM patients. This limitation hampers timely intervention and personalized treatment, highlighting the urgent need for improved risk prediction models and tailored management strategies.
This study aims to comprehensively investigate the association between the clinical status of T2DM and KPC by analyzing demographic and clinical characteristics in T2DM patients. A clinical prediction model was developed using multiple regression analysis to identify high-risk T2DM patients with KPC. This model is expected to provide clinicians with a practical tool to identify individuals at high risk of KPC, enabling targeted preventive measures to reduce infection rates and improve patient outcomes. In addition, this study aims to provide insights into the underlying mechanisms of T2DM and KPC, offering theoretical support for diabetes prevention and treatment strategies.
Materials and methods
Patient population
This retrospective study included 486 T2DM patients admitted to Shanghai Fifth People’s Hospital from December 2020 to December 2022.
Inclusion criteria: (1) Diagnosed with T2DM according to the Chinese Guidelines for the Prevention and Treatment of Type 2 Diabetes (2020) [10]; (2) Aged 18-75 years; (3) Absence of severe underlying diseases or cancer; (4) No obvious clinical symptoms of KP infection; (5) Availability of complete clinical data.
Exclusion criteria: (1) Pregnancy or nursing; (2) Use of antibiotics or immunomodulatory drugs within the past two weeks; (3) Presence of severe complications or comorbidities; (4) History of mental illness.
Fecal samples were collected from all participants for KP culture. Patients with positive culture results were assigned to the KPC group, while those with negative results were classified as the normal group. Ethical approval was obtained from the Ethics Committee of Shanghai Fifth People’s Hospital.
Data collection
Demographic and clinical data collected included: General demographics: age, sex, BMI, T2DM duration, history of hypertension, smoking, alcohol consumption, and hospitalization within six months. Clinical indicators: white blood cell count (WBC), neutrophil count (NC), hemoglobin (Hb), fasting blood glucose (FBG), alanine aminotransferase (ALT), aspartate aminotransferase (AST), creatinine (Cr), uric acid (UA), ferritin, fructosamine, thyroid-stimulating hormone (TSH), fasting insulin (FINS), and fasting C-peptide (FCP).
Statistical analysis
Quantitative variables were expressed as mean ± standard deviation and compared using independent samples t-tests. Qualitative variables were expressed as frequencies and percentages, with differences analyzed using Pearson’s chi-square test. Logistic regression analysis assessed the association between various factors and the risk of T2DM combined with KPC, providing odds ratios (ORs) and 95% confidence intervals (CIs). Variables with P < 0.05 in univariate logistic regression were included in the multivariate logistic regression analysis. Variance inflation factors (VIF) were calculated to assess collinearity in multivariate models. Statistical analyses were conducted using SPSS (version 27.0) and R (version 4.2.1). All tests were two-sided, with P < 0.05 considered statistically significant.
Model development and evaluation
The original dataset was randomly divided into a training set and an internal test set using a 7:3 ratio based on a random number method. Additionally, the case collection period was extended by six months to collect an external test set. A total of 96 patients were included in the external test set from January 2023 to June 2023. The training and test sets were assessed for balance. Logistic regression analysis was performed on the training set, and variables with P < 0.10 were included in the construction of the nomogram and the Receiver Operating Characteristic (ROC) curve.
The area under the ROC curve (AUC) was used to evaluate the discriminatory power of the prediction model. The Hosmer-Lemeshow goodness-of-fit test and calibration curve were employed to assess the model’s calibration. Decision curve analysis (DCA) was conducted, with treatment decision threshold probabilities as the horizontal axis and net benefit as the vertical axis, to evaluate the clinical utility of the prediction model across different thresholds. In addition, the accuracy, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated to comprehensively assess the model’s performance.
Results
General information
A total of 486 T2DM patients were included in the study. Based on KP test results, 362 patients were assigned to the normal group, while 124 patients were assigned to the KPC group, corresponding to a KPC rate of 25.51% (Table 1). The mean age of the normal and KPC groups was 62.66 ± 14.09 years and 62.83 ± 13.66 years, respectively, with no significant difference (P=0.905). The proportion of males was similar in both groups (55.25% vs. 54.03%, P=0.814). The mean BMI values for the normal and KPC groups were 24.70 ± 3.59 and 24.52 ± 3.70, respectively, with no significant difference (P=0.633) (Table 1).
Table 1.
Difference analysis between normal group and KPC group
| Variables | Total (n=486) | Normal group (n=362) | KPC group (n=124) | Statistic | P |
|---|---|---|---|---|---|
| Age, Mean ± SD | 62.70 ± 13.97 | 62.66 ± 14.09 | 62.83 ± 13.66 | t=-0.12 | 0.905 |
| Sex, n (%) | χ2=0.06 | 0.814 | |||
| Female | 219 (45.06) | 162 (44.75) | 57 (45.97) | ||
| Male | 267 (54.94) | 200 (55.25) | 67 (54.03) | ||
| BMI, Mean ± SD | 24.65 ± 3.62 | 24.70 ± 3.59 | 24.52 ± 3.70 | t=0.48 | 0.633 |
| Course of T2DM, Mean ± SD | 12.00 ± 8.34 | 11.58 ± 7.79 | 13.21 ± 9.72 | t=-1.69 | 0.093 |
| Hypertension, n (%) | χ2=0.31 | 0.579 | |||
| No | 226 (46.50) | 171 (47.24) | 55 (44.35) | ||
| Yes | 260 (53.50) | 191 (52.76) | 69 (55.65) | ||
| Smoking, n (%) | χ2=0.61 | 0.436 | |||
| No | 388 (79.84) | 286 (79.01) | 102 (82.26) | ||
| Yes | 98 (20.16) | 76 (20.99) | 22 (17.74) | ||
| Drinking, n (%) | χ2=1.01 | 0.315 | |||
| No | 449 (92.39) | 337 (93.09) | 112 (90.32) | ||
| Yes | 37 (7.61) | 25 (6.91) | 12 (9.68) | ||
| Hospitalized within six months, n (%) | χ2=11.50 | < 0.001 | |||
| No | 435 (89.51) | 334 (92.27) | 101 (81.45) | ||
| Yes | 51 (10.49) | 28 (7.73) | 23 (18.55) | ||
| WBC, Mean ± SD | 5.98 ± 1.90 | 5.79 ± 1.77 | 6.51 ± 2.13 | t=-3.38 | < 0.001 |
| NC, Mean ± SD | 3.54 ± 1.57 | 3.48 ± 1.47 | 3.72 ± 1.85 | t=-1.26 | 0.210 |
| Hb, Mean ± SD | 135.11 ± 16.75 | 136.22 ± 16.95 | 131.90 ± 15.75 | t=2.49 | 0.013 |
| FBG, Mean ± SD | 7.84 ± 3.18 | 7.72 ± 3.09 | 8.21 ± 3.39 | t=-1.52 | 0.130 |
| ALT, Mean ± SD | 21.88 ± 16.00 | 21.12 ± 16.20 | 24.09 ± 15.27 | t=-1.78 | 0.075 |
| AST, Mean ± SD | 19.24 ± 12.94 | 18.61 ± 13.11 | 21.07 ± 12.29 | t=-1.82 | 0.069 |
| Cr, Mean ± SD | 77.08 ± 29.78 | 78.32 ± 30.98 | 73.45 ± 25.75 | t=1.72 | 0.086 |
| UA, Mean ± SD | 305.84 ± 96.04 | 310.60 ± 100.37 | 291.97 ± 80.89 | t=2.08 | 0.039 |
| Ferritin, Mean ± SD | 302.15 ± 214.76 | 266.59 ± 196.37 | 405.96 ± 232.57 | t=-5.98 | < 0.001 |
| Fructosamine, Mean ± SD | 409.53 ± 109.78 | 412.72 ± 111.62 | 400.21 ± 104.09 | t=1.09 | 0.274 |
| TSH, Mean ± SD | 2.63 ± 2.06 | 2.51 ± 2.05 | 2.96 ± 2.05 | t=-2.06 | 0.040 |
| Fasting C-peptide, Mean ± SD | 0.63 ± 0.45 | 0.65 ± 0.47 | 0.59 ± 0.35 | t=1.26 | 0.209 |
Abbreviations: KPC, Klebsiella pneumoniae colonization; SD, standard deviation; BMI, body mass index; T2DM, type 2 diabetes mellitus; WBC, white blood cell; NC, neutrophil cell; Hb, hemoglobin; FBG, fasting blood glucose; ALT, alanine aminotransferase; AST, aspartate aminotransferase; Cr, creatinine; UA, uric acid; TSH, thyroid stimulating hormone.
Difference analysis between the two groups
Among the normal group, 28 patients (7.73%) had been hospitalized within six months, compared to 23 patients (18.55%) in the KPC group, showing a significant difference (P < 0.001) (Table 1). The WBC level was significantly higher in the KPC group (6.51 ± 2.13) compared to the normal group (5.79 ± 1.77, P < 0.001). The Hb level was lower in the KPC group (131.90 ± 15.75) than that in the normal group (136.22 ± 16.95, P=0.013). Additionally, significant differences were observed for UA (P=0.039), ferritin (P < 0.001), and TSH (P=0.040) between the two groups. However, no significant differences were found in T2DM duration, hypertension, smoking, alcohol consumption, neutrophil count, FBG, ALT, AST, Cr, fructosamine, or fasting C-peptide (all P > 0.05).
Logistic regression analysis
Univariate logistic regression identified several potential risk factors for KPC, including hospitalization within six months, elevated WBC, decreased Hb, elevated ferritin, and elevated TSH (Table 2). Multivariate logistic regression further confirmed that hospitalization within six months significantly increased the risk of KPC (OR=3.00, 95% CI: 1.54-5.82, P=0.001). Elevated WBC (OR=1.22, 95% CI: 1.08-1.38, P=0.001), decreased Hb (OR=0.98, 95% CI: 0.97-0.99, P=0.022), and elevated ferritin (OR=1.01, 95% CI: 1.01-1.01, P < 0.001) were also identified as independent risk factors.
Table 2.
Logistic regression analysis of T2DM patients with KPC
| Variables | Univariate Logistic regression | Multivariate Logistic regression | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
|
|
|
|||||||||
| β | S.E. | Z | P | OR (95% CI) | β | S.E. | Z | P | OR (95% CI) | |
| Hospitalized within six months | ||||||||||
| No | 1.00 (Reference) | 1.00 (Reference) | ||||||||
| Yes | 1.00 | 0.30 | 3.29 | < 0.001 | 2.72 (1.50-4.92) | 1.10 | 0.34 | 3.24 | 0.001 | 3.00 (1.54-5.82) |
| WBC | 0.20 | 0.06 | 3.60 | < 0.001 | 1.22 (1.10-1.37) | 0.20 | 0.06 | 3.20 | 0.001 | 1.22 (1.08-1.38) |
| Hb | -0.02 | 0.01 | -2.47 | 0.014 | 0.98 (0.97-0.99) | -0.02 | 0.01 | -2.29 | 0.022 | 0.98 (0.97-0.99) |
| Ferritin | 0.01 | 0.00 | 5.94 | < 0.001 | 1.01 (1.01-1.01) | 0.01 | 0.00 | 5.69 | < 0.001 | 1.01 (1.01-1.01) |
| TSH | 0.10 | 0.05 | 2.05 | 0.041 | 1.11 (1.01-1.22) | 0.09 | 0.06 | 1.68 | 0.093 | 1.10 (0.98-1.23) |
Abbreviations: KPC, Klebsiella pneumoniae colonization; T2DM, type 2 diabetes mellitus; OR, odds ratio; CI, confidence interval; WBC, white blood cell; NC, neutrophil cell; Hb, hemoglobin; TSH, thyroid stimulating hormone.
Construction of clinical prediction model
All participants were randomly divided into a training set and a test set. The clinical prediction model was developed based on the results of multivariate logistic regression analysis using the training set data. The model included the following predictors: hospitalization within six months, WBC, Hb, ALT, AST, and ferritin. These predictors were visualized in a nomogram (Figure 1).
Figure 1.
A nomogram of the clinical predictive model.
Balance check between training and validation sets
To ensure the statistical comparability of the training and test sets, the raw data were split into a 7:3 ratio, and the two groups were compared for differences in basic and clinical features. The results showed no significant differences (P > 0.05), confirming that the two groups were balanced (Table 3).
Table 3.
Balance check between training and testing sets
| Variables | Total (n=486) | test (n=146) | train (n=340) | Statistic | P |
|---|---|---|---|---|---|
| Age, Mean ± SD | 62.70 ± 13.97 | 61.71 ± 14.85 | 63.13 ± 13.58 | t=-1.03 | 0.303 |
| Sex, n (%) | χ2=0.00 | 0.967 | |||
| No | 219 (45.06) | 66 (45.21) | 153 (45.00) | ||
| Yes | 267 (54.94) | 80 (54.79) | 187 (55.00) | ||
| BMI, Mean ± SD | 24.65 ± 3.62 | 24.83 ± 3.82 | 24.57 ± 3.53 | t=0.72 | 0.470 |
| Course of T2DM, Mean ± SD | 12.00 ± 8.34 | 12.09 ± 8.64 | 11.96 ± 8.22 | t=0.16 | 0.875 |
| Hypertension, n (%) | χ2=0.00 | 0.983 | |||
| No | 226 (46.50) | 68 (46.58) | 158 (46.47) | ||
| Yes | 260 (53.50) | 78 (53.42) | 182 (53.53) | ||
| Smoking, n (%) | χ2=0.72 | 0.396 | |||
| No | 388 (79.84) | 120 (82.19) | 268 (78.82) | ||
| Yes | 98 (20.16) | 26 (17.81) | 72 (21.18) | ||
| Drinking, n (%) | χ2=0.17 | 0.677 | |||
| No | 449 (92.39) | 136 (93.15) | 313 (92.06) | ||
| Yes | 37 (7.61) | 10 (6.85) | 27 (7.94) | ||
| Hospitalized within six months, n (%) | χ2=1.41 | 0.235 | |||
| No | 435 (89.51) | 127 (86.99) | 308 (90.59) | ||
| Yes | 51 (10.49) | 19 (13.01) | 32 (9.41) | ||
| WBC, Mean ± SD | 5.98 ± 1.90 | 5.97 ± 1.93 | 5.98 ± 1.89 | t=-0.08 | 0.936 |
| NC, Mean ± SD | 3.54 ± 1.57 | 3.54 ± 1.60 | 3.55 ± 1.56 | t=-0.04 | 0.968 |
| Hb, Mean ± SD | 135.11 ± 16.75 | 136.87 ± 16.67 | 134.36 ± 16.75 | t=1.52 | 0.130 |
| FBG, Mean ± SD | 7.84 ± 3.18 | 7.57 ± 2.99 | 7.96 ± 3.25 | t=-1.22 | 0.221 |
| ALT, Mean ± SD | 21.88 ± 16.00 | 22.48 ± 15.57 | 21.62 ± 16.20 | t=0.55 | 0.586 |
| AST, Mean ± SD | 19.24 ± 12.94 | 18.94 ± 13.00 | 19.37 ± 12.93 | t=-0.33 | 0.739 |
| Cr, Mean ± SD | 77.08 ± 29.78 | 76.10 ± 30.18 | 77.50 ± 29.64 | t=-0.48 | 0.633 |
| UA, Mean ± SD | 305.84 ± 96.04 | 293.68 ± 92.92 | 311.06 ± 97.02 | t=-1.83 | 0.067 |
| Ferritin, Mean ± SD | 302.15 ± 214.76 | 310.34 ± 211.84 | 298.64 ± 216.21 | t=0.55 | 0.582 |
| Fructosamine, Mean ± SD | 409.53 ± 109.78 | 418.63 ± 103.60 | 405.62 ± 112.24 | t=1.20 | 0.231 |
| TSH, Mean ± SD | 2.63 ± 2.06 | 2.74 ± 2.00 | 2.58 ± 2.09 | t=0.77 | 0.444 |
| Fasting C-peptide, Mean ± SD | 0.63 ± 0.45 | 0.62 ± 0.45 | 0.64 ± 0.45 | t=-0.46 | 0.642 |
Abbreviations: KPC, Klebsiella pneumoniae colonization; SD, standard deviation; BMI, body mass index; T2DM, type 2 diabetes mellitus; WBC, white blood cell; NC, neutrophil cell; Hb, hemoglobin; FBG, fasting blood glucose; ALT, alanine aminotransferase; AST, aspartate aminotransferase; Cr, creatinine; UA, uric acid; TSH, thyroid stimulating hormone.
Difference analysis and regression analysis of training set
In the training set, significant differences between the normal and KPC groups were observed for hospitalization within six months (P=0.001), WBC (P=0.029), ALT (P=0.045) and ferritin (P < 0.001) (Table 4). Regression analysis revealed that the following factors were associated with an increased risk of KPC: Hospitalization within six months (OR=4.26, 95% CI: 1.91-9.48, P < 0.001); High WBC levels (OR=1.19, 95% CI: 1.03-1.37, P=0.015); Elevated ALT levels (OR=1.02, 95% CI: 1.01-1.04, P=0.012); Elevated ferritin levels (OR=1.01, 95% CI: 1.01-1.01, P < 0.001) (Table 5).
Table 4.
Differential analysis of training set
| Variables | Total (n=340) | Normal group (n=252) | KPC group (n=88) | Statistic | P |
|---|---|---|---|---|---|
| Age, Mean ± SD | 63.13 ± 13.58 | 63.54 ± 13.85 | 61.95 ± 12.77 | t=0.94 | 0.346 |
| Sex, n (%) | χ2=0.02 | 0.881 | |||
| Female | 153 (45.00) | 114 (45.24) | 39 (44.32) | ||
| Male | 187 (55.00) | 138 (54.76) | 49 (55.68) | ||
| BMI, Mean ± SD | 24.57 ± 3.53 | 24.63 ± 3.52 | 24.39 ± 3.59 | t=0.55 | 0.582 |
| Course of T2DM, Mean ± SD | 11.96 ± 8.22 | 11.88 ± 7.79 | 12.18 ± 9.39 | t=-0.27 | 0.788 |
| Hypertension, n (%) | χ2=0.05 | 0.824 | |||
| No | 158 (46.47) | 118 (46.83) | 40 (45.45) | ||
| Yes | 182 (53.53) | 134 (53.17) | 48 (54.55) | ||
| Smoking, n (%) | χ2=1.21 | 0.271 | |||
| No | 268 (78.82) | 195 (77.38) | 73 (82.95) | ||
| Yes | 72 (21.18) | 57 (22.62) | 15 (17.05) | ||
| Drinking, n (%) | χ2=0.85 | 0.357 | |||
| No | 313 (92.06) | 234 (92.86) | 79 (89.77) | ||
| Yes | 27 (7.94) | 18 (7.14) | 9 (10.23) | ||
| Hospitalized within six months, n (%) | χ2=10.71 | 0.001 | |||
| No | 308 (90.59) | 236 (93.65) | 72 (81.82) | ||
| Yes | 32 (9.41) | 16 (6.35) | 16 (18.18) | ||
| WBC, Mean ± SD | 5.98 ± 1.89 | 5.85 ± 1.79 | 6.36 ± 2.11 | t=-2.19 | 0.029 |
| NC, Mean ± SD | 3.55 ± 1.56 | 3.57 ± 1.49 | 3.48 ± 1.75 | t=0.41 | 0.682 |
| Hb, Mean ± SD | 134.36 ± 16.75 | 135.32 ± 16.96 | 131.61 ± 15.90 | t=1.79 | 0.074 |
| FBG, Mean ± SD | 7.96 ± 3.25 | 7.85 ± 3.20 | 8.28 ± 3.39 | t=-1.07 | 0.285 |
| ALT, Mean ± SD | 21.62 ± 16.20 | 20.58 ± 16.41 | 24.60 ± 15.28 | t=-2.01 | 0.045 |
| AST, Mean ± SD | 19.37 ± 12.93 | 18.62 ± 13.07 | 21.51 ± 12.33 | t=-1.81 | 0.071 |
| Cr, Mean ± SD | 77.50 ± 29.64 | 78.63 ± 30.86 | 74.29 ± 25.76 | t=1.18 | 0.238 |
| UA, Mean ± SD | 311.06 ± 97.02 | 316.30 ± 100.35 | 296.08 ± 85.53 | t=1.69 | 0.092 |
| Ferritin, Mean ± SD | 298.64 ± 216.21 | 268.05 ± 195.98 | 386.23 ± 246.73 | t=-4.07 | < 0.001 |
| Fructosamine, Mean ± SD | 405.62 ± 112.24 | 409.35 ± 115.91 | 394.94 ± 100.86 | t=1.04 | 0.301 |
| TSH, Mean ± SD | 2.58 ± 2.09 | 2.47 ± 2.10 | 2.88 ± 2.04 | t=-1.58 | 0.115 |
| Fasting C-peptide, Mean ± SD | 0.64 ± 0.45 | 0.65 ± 0.47 | 0.60 ± 0.37 | t=1.09 | 0.276 |
Abbreviations: KPC, Klebsiella pneumoniae colonization; SD, standard deviation; BMI, body mass index; T2DM, type 2 diabetes mellitus; WBC, white blood cell; NC, neutrophil cell; Hb, hemoglobin; FBG, fasting blood glucose; ALT, alanine aminotransferase; AST, aspartate aminotransferase; Cr, creatinine; UA, uric acid; TSH, thyroid stimulating hormone.
Table 5.
Logistic regression analysis of training set
| Variables | Univariate Logistic regression | Multivariate Logistic regression | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
|
||||||||||
| β | S.E. | Z | P | OR (95% CI) | β | S.E. | Z | P | OR (95% CI) | VIF | |
| Hospitalized within six months | |||||||||||
| No | 1.00 (Reference) | 1.00 (Reference) | |||||||||
| Yes | 1.19 | 0.38 | 3.14 | 0.002 | 3.28 (1.56-6.88) | 1.45 | 0.41 | 3.54 | < 0.001 | 4.26 (1.91-9.48) | 1.006 |
| WBC | 0.14 | 0.07 | 2.17 | 0.030 | 1.15 (1.01-1.31) | 0.18 | 0.07 | 2.44 | 0.015 | 1.19 (1.03-1.37) | 1.039 |
| Hb | -0.01 | 0.01 | -1.78 | 0.075 | 0.99 (0.97-1.00) | -0.01 | 0.01 | -1.41 | 0.158 | 0.99 (0.97-1.00) | 1.013 |
| ALT | 0.01 | 0.01 | 1.99 | 0.046 | 1.02 (1.01-1.03) | 0.02 | 0.01 | 2.52 | 0.012 | 1.02 (1.01-1.04) | 1.031 |
| AST | 0.02 | 0.01 | 1.80 | 0.072 | 1.02 (1.00-1.04) | 0.01 | 0.01 | 1.19 | 0.232 | 1.01 (0.99-1.03) | 1.024 |
| Ferritin | 0.01 | 0.00 | 4.26 | < 0.001 | 1.01 (1.01-1.01) | 0.01 | 0.00 | 4.27 | < 0.001 | 1.01 (1.01-1.01) | 1.020 |
Abbreviations: OR, odds ratio; CI, confidence interval; VIF, variance inflation factor; WBC, white blood cell; Hb, hemoglobin; ALT, alanine aminotransferase; AST, aspartate aminotransferase.
Model evaluation
The ROC curve of the prediction model yielded an AUC value of 0.74 (95% CI: 0.68-0.80) (Figure 2A), indicating good discriminatory power. The Hosmer-Lemeshow goodness-of-fit test (P=0.650) and the calibration curve showed no significant differences between observed and predicted values, confirming good model calibration (Figure 2D). Decision curve analysis (DCA) demonstrated that the model provided a positive net benefit for patients within a threshold probability range of 0.1-1.0 (Figure 2G).
Figure 2.
ROC curve, calibration curve, DCA curve of training set and test set. (A, D, G) ROC curve, calibration curve and DCA curve of the training set; (B, E, H) ROC curve, calibration curve and DCA curve of the internal test set. (C, F, I) ROC curve, calibration curve and DCA curve of the external testing set.
Model validation
The internal validation set included 146 T2DM patients, of whom 36 were KPC-positive, resulting in a KPC rate of 24.66%. The AUC of the internal test set was 0.76 (95% CI: 0.66-0.86) (Figure 2B). The Hosmer-Lemeshow goodness-of-fit test (P=0.290) and the calibration curve results showed that the model fits well (Figure 2E). DCA indicated a positive net benefit within the probability range of 0.1-1.0 (Figure 2H).
The external test set included 96 patients, of whom 25 were KPC-positive, corresponding to a KPC rate of 26.04%. The external test set’s AUC was 0.75 (95% CI: 0.62-0.87) (Figure 2C). The Hosmer-Lemeshow goodness-of-fit test result was 0.078 (Figure 2F). The DCA results showed that the model provided good net returns within the threshold probability range of 0.2-1.0 (Figure 2I).
Confusion matrix analysis of the training, internal test, and external test sets
In the training set, the clinical prediction model demonstrated an accuracy of 0.65 (95% CI: 0.59-0.70). Sensitivity and specificity were 0.59 (95% CI: 0.53-0.65) and 0.82 (95% CI: 0.74-0.90), respectively. Positive predictive value (PPV) and negative predictive value (NPV) were 0.90 (95% CI: 0.86-0.95) and 0.41 (95% CI: 0.34-0.48), respectively (Table 6).
Table 6.
Analysis of confusion matrix between training set and test set
| Data | AUC (95% CI) | Accuracy (95% CI) | Sensitivity (95% CI) | Specificity (95% CI) | PPV (95% CI) | NPV (95% CI) | Cut off |
|---|---|---|---|---|---|---|---|
| Training set | 0.74 (0.68-0.80) | 0.65 (0.59-0.70) | 0.59 (0.53-0.65) | 0.82 (0.74-0.90) | 0.90 (0.86-0.95) | 0.41 (0.34-0.48) | 0.206 |
| Internal test set | 0.76 (0.66-0.86) | 0.63 (0.55-0.71) | 0.57 (0.48-0.67) | 0.81 (0.68-0.93) | 0.90 (0.83 - 0.97) | 0.38 (0.27-0.49) | 0.206 |
| External test set | 0.75 (0.62-0.87) | 0.80 (0.71-0.88) | 0.85 (0.76-0.93) | 0.68 (0.50-0.86) | 0.88 (0.81-0.96) | 0.61 (0.43-0.79) | 0.299 |
Abbreviations: AUC, area under the curve; CI, confidence interval; PPV, positive predictive value; NPV, negative predictive value.
In the internal test set, the accuracy was 0.63 (95% CI: 0.55-0.71). Sensitivity and specificity were 0.57 (95% CI: 0.48-0.67) and 0.81 (95% CI: 0.68-0.93), respectively. PPV and NPV were 0.90 (95% CI: 0.83-0.97) and 0.38 (95% CI: 0.27-0.49), respectively.
For the external test set, the model exhibited an accuracy of 0.80 (95% CI: 0.71-0.88). Sensitivity, specificity, PPV, and NPV were 0.85 (95% CI: 0.76-0.93), 0.68 (95% CI: 0.50-0.86), 0.88 (95% CI: 0.81-0.96), and 0.61 (95% CI: 0.43-0.79), respectively (Table 6).
Discussion
T2DM is a chronic metabolic disease characterized by insulin resistance and impaired insulin secretion, leading to persistent hyperglycemia [11]. T2DM accounts for over 90% of diabetes cases and damages multiple organs, including the kidneys, cardiovascular system, nervous system, and eyes. It also increases susceptibility to infections [12]. Hyperglycemia impairs immune defenses by reducing alveolar macrophages’ recognition and adhesion abilities and restricting lysozyme synthesis, which weakens antibacterial functions [13]. Additionally, hyperglycemia compromises immune cell phagocytosis, further reducing infection resistance [14].
This study identified several risk factors associated with KPC in T2DM patients: hospitalization within the past six months, elevated WBC levels, reduced Hb levels, and increased ferritin and TSH levels. A clinical prediction model was constructed based on these factors and demonstrated good predictive performance.
Recent hospitalization is a well-recognized risk factor for healthcare-associated infections due to increased exposure to resistant pathogens [15]. Reports indicate that KP infections account for approximately one-third of Gram-negative bacterial infections in hospitalized patients, with a high risk of developing drug-resistant strains [16,17]. In T2DM patients, the microenvironmental changes, including acidic tissue fluids and nutrient abundance, facilitate KP colonization. Once colonized, KP may disseminate via the bloodstream, causing complications like sepsis and endophthalmitis [9]. Hospitalization also disrupts normal microbiota and often involves exposure to broad-spectrum antibiotics, further increasing the likelihood of KPC in T2DM patients [18].
Elevated WBC levels in T2DM patients were associated with increased susceptibility to KPC. As a marker of immune response, high WBC levels often reflect infection [19]. However, chronic inflammation - a hallmark of T2DM - can impair immune system function, reducing its ability to combat infections [11]. Elevated WBC levels may thus represent both immune system activation due to KP infection and an underlying dysfunction, increasing the risk of KPC [20].
This study found that decreased Hb levels correlated with a higher risk of KPC. Diabetic nephropathy, a common microvascular complication of T2DM, damages kidney function, leading to reduced erythropoietin secretion and subsequent anemia [21,22]. Lower Hb levels may signal advanced T2DM, which is more susceptible to bacterial infections. While the exact mechanism linking Hb and KPC requires further investigation, this finding suggests that anemia could be a marker for heightened vulnerability to KPC.
This study identified elevated ferritin as a factor associated with increased susceptibility to KPC in T2DM patients. The underlying mechanism may involve multiple pathways and complex interactions. Iron overload, for example, can enhance infection risk by providing bacteria with the iron required for growth [23]. Elevated ferritin levels are not only indicative of iron deficiency anemia but are also linked to various inflammatory, infectious, and oncologic conditions [24]. Ferritin elevation may result in increased free iron availability, facilitating KP growth [25,26]. As an acute-phase protein, elevated ferritin levels may reflect enhanced inflammatory responses, which could promote KPC [27]. Furthermore, dysregulated iron metabolism may impair immune function, exacerbating infection risk [28]. For T2DM patients, monitoring and regulating ferritin levels could represent a novel strategy for preventing KP infections.
This study also highlighted the potential roles of TSH and liver function (ALT) in KPC susceptibility. Elevated TSH levels may impair immune cell activity, making it more difficult for T2DM patients to eliminate KP, thereby increasing the risk of colonization [29]. Similarly, elevated ALT levels may indicate liver inflammation or injury, which could compromise detoxification and metabolic functions, ultimately affecting immune responses and antibacterial capacity [30]. Increased KPC susceptibility in T2DM patients likely involves multiple mechanisms, including inflammatory reactions, immune dysfunction, and dysbiosis exacerbated by hyperglycemia. Metabolic abnormalities such as anemia and iron overload may also provide favorable conditions for pathogen growth. These findings offer new insights into the relationship between T2DM and KP colonization and underscore the need to focus on these indicators in future research on bacterial infections.
This study has certain limitations. First, as a retrospective analysis, it relies on previously collected data, which may introduce selection and recall biases. Second, the single-center nature of the study limits the generalizability of its findings. Third, the heterogeneity of T2DM manifestations and the multifactorial nature of KPC underscore the need for cautious interpretation of the results.
Despite these constraints, this study constructed a clinical prediction model for identifying high-risk T2DM patients prone to KPC. This model enables early identification and targeted intervention, such as enhanced surveillance, optimized glucose control, and adjusted antibiotic strategies, reducing infection risks and improving patient outcomes. Moreover, it facilitates the development of individualized treatment plans, enhancing therapeutic effectiveness while minimizing unnecessary medical resource utilization.
In conclusion, this study investigated the association between the basic and clinical characteristics of T2DM patients and their susceptibility to KPC through retrospective analysis. The findings identified hospitalization within the past six months, elevated WBC, decreased Hb, and elevated ferritin as independent risk factors for KPC. Additionally, thyroid and liver function indicators may influence susceptibility.
The clinical prediction model developed in this study demonstrated strong predictive capability in identifying high-risk populations for KPC. These findings provide theoretical insights into the clinical characteristics of T2DM and its relationship with KPC. They can aid clinicians in early identification and intervention for high-risk T2DM patients, enabling personalized treatment and improving patient prognosis.
Acknowledgements
This work was supported by Scientific Research Project funded by Shanghai Fifth People’s Hospital, Fudan University (No. 2020WYFY02)and Talent Development Plan funded by Shanghai Fifth People’s Hospital, Fudan University (No. 2024WYRCSG07).
Disclosure of conflict of interest
None.
References
- 1.Chen Y, Zhao J, Li X, Xie Z, Huang G, Yan X, Zhou H, Zheng L, Xu T, Zhou K, Zhou Z. Prevalence of maturity-onset diabetes of the young in phenotypic type 2 diabetes in young adults: a nationwide, multi-center, cross-sectional survey in China. Chin Med J (Engl) 2023;136:56–64. doi: 10.1097/CM9.0000000000002321. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Xie J, Wang M, Long Z, Ning H, Li J, Cao Y, Liao Y, Liu G, Wang F, Pan A. Global burden of type 2 diabetes in adolescents and young adults, 1990-2019: systematic analysis of the Global Burden of Disease Study 2019. BMJ. 2022;379:e072385. doi: 10.1136/bmj-2022-072385. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kautzky-Willer A, Harreiter J, Pacini G. Sex and gender differences in risk, pathophysiology and complications of type 2 diabetes mellitus. Endocr Rev. 2016;37:278–316. doi: 10.1210/er.2015-1137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Yun JS, Ko SH. Current trends in epidemiology of cardiovascular disease and cardiovascular risk management in type 2 diabetes. Metabolism. 2021;123:154838. doi: 10.1016/j.metabol.2021.154838. [DOI] [PubMed] [Google Scholar]
- 5.Ali MK, Pearson-Stuttard J, Selvin E, Gregg EW. Interpreting global trends in type 2 diabetes complications and mortality. Diabetologia. 2022;65:3–13. doi: 10.1007/s00125-021-05585-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Stojowska-Swędrzyńska K, Łupkowska A, Kuczyńska-Wiśnik D, Laskowska E. Antibiotic Heteroresistance in Klebsiella pneumoniae. Int J Mol Sci. 2021;23:449. doi: 10.3390/ijms23010449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Namikawa H, Oinuma KI, Yamada K, Kaneko Y, Kakeya H, Shuto T. Predictors of hypervirulent Klebsiella pneumoniae infections: a systematic review and meta-analysis. J Hosp Infect. 2023;134:153–160. doi: 10.1016/j.jhin.2023.02.005. [DOI] [PubMed] [Google Scholar]
- 8.Chen S, Tan X, Li X, Zhao T. Association between the duration of diabetes and gram-negative bacterial infection in diabetic foot infections: a case-control study. Endocr J. 2022;69:1061–1065. doi: 10.1507/endocrj.EJ21-0690. [DOI] [PubMed] [Google Scholar]
- 9.Martin RM, Bachman MA. Colonization, infection, and the accessory genome of klebsiella pneumoniae. Front Cell Infect Microbiol. 2018;8:4. doi: 10.3389/fcimb.2018.00004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Chinese Diabetes Society. The Chinese type 2 diabetes prevention guide (2020 edition) (on) Chinese Journal of Practical Internal Medicine. 2021;9:668–695. [Google Scholar]
- 11.Ahmad E, Lim S, Lamptey R, Webb DR, Davies MJ. Type 2 diabetes. Lancet. 2022;400:1803–1820. doi: 10.1016/S0140-6736(22)01655-5. [DOI] [PubMed] [Google Scholar]
- 12.Demir S, Nawroth PP, Herzig S, Ekim Üstünel B. Emerging targets in type 2 diabetes and diabetic complications. Adv Sci (Weinh) 2021;8:e2100275. doi: 10.1002/advs.202100275. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Mo Y, Mo L, Zhang Y, Zhang Y, Yuan J, Zhang Q. High glucose enhances the activation of NLRP3 inflammasome by ambient fine particulate matter in alveolar macrophages. Part Fibre Toxicol. 2023;20:41. doi: 10.1186/s12989-023-00552-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Huang YT, Jiang JY, Hsu MS, Hsu HS, Liao CH, Hsueh PR. The prevalence of rectal carriage of Klebsiella pneumoniae amongst diabetic patients and their clinical relevance in Taiwan: a five-year prospective study. J Microbiol Immunol Infect. 2018;51:510–518. doi: 10.1016/j.jmii.2017.05.009. [DOI] [PubMed] [Google Scholar]
- 15.Pu D, Zhao J, Chang K, Zhuo X, Cao B. “Superbugs” with hypervirulence and carbapenem resistance in Klebsiella pneumoniae: the rise of such emerging nosocomial pathogens in China. Sci Bull (Beijing) 2023;68:2658–2670. doi: 10.1016/j.scib.2023.09.040. [DOI] [PubMed] [Google Scholar]
- 16.Gorrie CL, Mirčeta M, Wick RR, Judd LM, Lam MMC, Gomi R, Abbott IJ, Thomson NR, Strugnell RA, Pratt NF, Garlick JS, Watson KM, Hunter PC, Pilcher DV, McGloughlin SA, Spelman DW, Wyres KL, Jenney AWJ, Holt KE. Genomic dissection of Klebsiella pneumoniae infections in hospital patients reveals insights into an opportunistic pathogen. Nat Commun. 2022;13:3017. doi: 10.1038/s41467-022-30717-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Liu YN, Zhang YF, Xu Q, Qiu Y, Lu QB, Wang T, Zhang XA, Lin SH, Lv CL, Jiang BG, Li H, Li ZJ, Gao GF, Yang WZ, Hay SI, Wang LP, Fang LQ, Liu W Chinese Center for Disease Control and Prevention Etiology Surveillance Study Team of Acute Respiratory Infections. Infection and co-infection patterns of community-acquired pneumonia in patients of different ages in China from 2009 to 2020: a national surveillance study. Lancet Microbe. 2023;4:e330–e339. doi: 10.1016/S2666-5247(23)00031-9. [DOI] [PubMed] [Google Scholar]
- 18.Bantar C, Alcazar G, Franco D, Vesco E, Salamone F, Izaguirre M, Oliva ME. Impact of antibiotic treatment on bacterial resistance rates from patients with hospital-acquired infection. J Chemother. 2007;19:673–676. doi: 10.1179/joc.2007.19.6.673. [DOI] [PubMed] [Google Scholar]
- 19.Ratter-Rieck JM, Maalmi H, Trenkamp S, Zaharia OP, Rathmann W, Schloot NC, Straßburger K, Szendroedi J, Herder C, Roden M German Diabetes Study (GDS) Group; GDS Group. Leukocyte counts and t-cell frequencies differ between novel subgroups of diabetes and are associated with metabolic parameters and biomarkers of inflammation. Diabetes. 2021;70:2652–2662. doi: 10.2337/db21-0364. [DOI] [PubMed] [Google Scholar]
- 20.Daryabor G, Atashzar MR, Kabelitz D, Meri S, Kalantar K. The effects of type 2 diabetes mellitus on organ metabolism and the immune system. Front Immunol. 2020;11:1582. doi: 10.3389/fimmu.2020.01582. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Rayego-Mateos S, Rodrigues-Diez RR, Fernandez-Fernandez B, Mora-Fernández C, Marchant V, Donate-Correa J, Navarro-González JF, Ortiz A, Ruiz-Ortega M. Targeting inflammation to treat diabetic kidney disease: the road to 2030. Kidney Int. 2023;103:282–296. doi: 10.1016/j.kint.2022.10.030. [DOI] [PubMed] [Google Scholar]
- 22.Tonneijck L, Muskiet MH, Smits MM, van Bommel EJ, Heerspink HJ, van Raalte DH, Joles JA. Glomerular hyperfiltration in diabetes: mechanisms, clinical significance, and treatment. J Am Soc Nephrol. 2017;28:1023–1039. doi: 10.1681/ASN.2016060666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Chang X, Lv C, Zhao G. A dual function of ferritin (animal and plant): its holo form for iron supplementation and apo form for delivery systems. Annu Rev Food Sci Technol. 2023;14:113–133. doi: 10.1146/annurev-food-060721-024902. [DOI] [PubMed] [Google Scholar]
- 24.Israel A, Bornstein G, Gilad L, Shechtman L, Furie N, Ben-Zvi I, Grossman C. Clinical and prognostic significance of elevated ferritin levels in hospitalised adults. Postgrad Med J. 2022;98:622–625. doi: 10.1136/postgradmedj-2021-139832. [DOI] [PubMed] [Google Scholar]
- 25.Chen T, Dong G, Zhang S, Zhang X, Zhao Y, Cao J, Zhou T, Wu Q. Effects of iron on the growth, biofilm formation and virulence of Klebsiella pneumoniae causing liver abscess. BMC Microbiol. 2020;20:36. doi: 10.1186/s12866-020-01727-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Symeonidis AS. The role of iron and iron chelators in zygomycosis. Clin Microbiol Infect. 2009;15(Suppl 5):26–32. doi: 10.1111/j.1469-0691.2009.02976.x. [DOI] [PubMed] [Google Scholar]
- 27.Gehrer CM, Mitterstiller AM, Grubwieser P, Meyron-Holtz EG, Weiss G, Nairz M. Advances in ferritin physiology and possible implications in bacterial infection. Int J Mol Sci. 2023;24:4659. doi: 10.3390/ijms24054659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Roemhild K, von Maltzahn F, Weiskirchen R, Knüchel R, von Stillfried S, Lammers T. Iron metabolism: pathophysiology and pharmacology. Trends Pharmacol Sci. 2021;42:640–656. doi: 10.1016/j.tips.2021.05.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Cui C, Sui H, Wang Z, Zhang T, Zheng J, Yan H, Li Q, Mo Z, Liu L. Thyroid hormone sensitivity and diabetes onset: a longitudinal cross-lagged cohort. Front Endocrinol (Lausanne) 2023;14:1267612. doi: 10.3389/fendo.2023.1267612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Kesenogile B, Godman B, Rwegerera GM. Alanine transaminase and hemoglobin appear to predict the occurrence of antituberculosis medication hepatotoxicity; findings and implications in Botswana. Expert Rev Anti Infect Ther. 2021;19:379–391. doi: 10.1080/14787210.2020.1822735. [DOI] [PubMed] [Google Scholar]