HbA1c is Positively Associated with Serum Carcinoembryonic Antigen (CEA) in Patients with Diabetes: A Cross-Sectional Study.

INTRODUCTION: To explore the influence of glycemic levels on the serum carcinoembryonic antigen (CEA) levels in patients with diabetes.
METHODS: This is a cross-sectional study. Patients with diabetes aged 18-75 years old were recruited. Those patients with carcinoma, abnormal renal function (serum creatinine ≥ 115 μmol/l), and abnormal hepatic function (serum alanine aminotransferase ≥ 97.5 U/l) were excluded in this study. Blood samples were obtained from every patient after an overnight fasting, and CEA was determined using a chemiluminescent particle immunoassay.
RESULTS: A total of 324 patients with type 1 diabetes and 3019 patients with type 2 diabetes were included in this study. Patients with type 1 diabetes had higher levels of HbA1c (9.5% vs. 8.7%) and CEA (2.79 vs. 2.34 ng/ml), but lower fasting C peptide (0.72 vs. 1.71 ng/ml) than those with type 2 diabetes (all P < 0.001). Data indicated that CEA was higher in patients with smoking, drinking, older age, higher levels of HbA1c, and lower level of fasting C peptide (all P < 0.05). Multiple linear regression analysis indicated that CEA was independently associated with smoking, age, BMI, and HbA1c (all P < 0.05).
CONCLUSION: HbA1c was positively associated with the levels of CEA in patients with diabetes. More studies are warranted to investigate whether elevated CEA is associated with increased carcinoma risk in patients with diabetes.

Introduction

Carcinoembryonic antigen (CEA) has been always used as a tumor marker for diagnosis and predicting the prognosis of carcinoma of the colorectum, lung, thyroid, etc. [1-3]. CEA was monitored in patients with carcinoma in order to judge the disease process and give clues to recurrence in those under radical resection [4-6]. Moreover, elevated CEA levels were also found in patients with carcinoma of the stomach, liver, lung, pancreas, breast, ovary, uterus/cervix uteri, and urinary organs [5, 7–12]. CEA may also increase in such pathological conditions as hepatic cirrhosis/inflammation, pulmonary emphysema, rectal polyps, and colon inflammation [13, 14].

Diabetes, especially type 2 diabetes, was associated with increased risk incidence of carcinoma. A prospective study in China indicated that patients with type 2 diabetes mellitus were associated with higher incidence and mortality risks of cancer, especially pancreatic cancer, in comparison to the general population [15]. The Veteran Administration Registry Study also reported an increasing incidence of pancreatic carcinoma in patients with type 2 diabetes in comparison to the non-diabetic controls [16]. It has also been reported that uncontrolled severe hyperglycemia rather than preoperative diabetes mellitus negatively affects the survival outcomes following pancreatic ductal adenocarcinoma (PDAC) resection [17]. However, the level of CEA was not investigated in this study despite higher levels of CEA being recognized as a tumor marker in these tumors.

Several studies indicated that the status of diabetes affects serous levels of markers of carcinoma. The level of CA 19-9 has been explored in various studies; it was increased in patients with type 2 diabetes and highly correlated with glycemic control [18-20]. CEA has also been reported to be associated with hyperglycemia in patients with diabetes in small-scale studies [19, 20]. However, the influence of glycemic control on the levels of CEA was scarcely investigated in large-scale samples. In this study, we intended to investigate the factors affecting the level of CEA in Chinese patients with diabetes.

Methods

Subjects

This study was launched from March 2012 to December 2016 and performed at the Department of Endocrinology and Metabolism, Shanghai University of Medicine and Health Sciences Affiliated Fengxian Hospital in Shanghai, China. The inclusion criteria included age between 18 and 75; with complete clinical data. Those patients with malignancy (n = 8), renal dysfunction (serum creatinine ≥ 115 μmol/l, n = 111), and abnormal hepatic function (serum alanine aminotransferase ≥ 97.5 U/l, n = 64) were excluded according to the study protocol. In total, 3343 patients with diabetes aged 18–75 years were included in this study. Data on demographic information and clinical parameters were collected via patients’ records.

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1964, as revised in 2013. This study was approved by the Shanghai University of Medicine and Health Sciences Affiliated Fengxian Hospital institutional review board. All patients gave written informed consent prior to data collection.

Anthropometric and Biochemical Measurements

Body mass index (BMI) was assessed with the formula as follows: BMI (kg/m2) = body weight (kg)/height squared (m2). Height and weight were measured according to the standard protocol.

Subjects fasted overnight for at least 10 h before venous blood samples were drawn. Parameters such as glycosylated hemoglobin (HbA1c), fasting plasma glucose (FPG), lipid profiles, and fasting C peptide (FCP) were determined with the methods as previously described [21]. CEA was determined using a chemiluminescent particle immunoassay (ARCHITECT, Abbott Ireland Diagnostics Division, Sligo, Ireland).

Definition of Diabetes Types

The World Health Organization (WHO) report (1999) defined diabetes as FPG ≥ 7.0 mmol/l and/or a 2-h post-load plasma glucose (2hPG) ≥ 11.1 mmol/l, or being on medical treatment for diabetes [22]. Type 1 diabetes mellitus and type 2 diabetes mellitus were identified according to the criteria established by the American Diabetes Association (2010) [23]. Insulin deficiency was defined as FCP < 0.9 ng/ml.

Statistical Analysis

Continuous variables were presented as median (interquartile range, IQR), and categorical variables as number (percentage). Differences in medians were assessed using the Mann–Whitney U test between two groups and the Kruskal–Wallis test among three or four groups; differences in proportions were analyzed using the Chi square test. The correlation between CEA and clinical characteristics was tested using multiple linear regression analysis by backward step method. The age/HbA1c-fitted CEA levels were calculated using Stata software, all other statistical analyses were performed using SPSS 19.0 (SPSS Inc., Chicago, IL, USA); two-sided P values less than 0.05 were considered statistically significant.

Results

Clinical Characteristics of the Patients

As shown in Table 1, patients with type 1 diabetes had shorter duration of diabetes, but higher levels of HbA1c, high density lipoprotein-cholesterol (HDL-C) and CEA, but lower BMI, FCP, alanine aminotransferase (ALT), and triglyceride (TG). Table 2 shows the clinical characteristics in patients with different CEA quartiles. There were significant differences in gender composition, proportion of current smoker and drinker, age, duration of diabetes, BMI, and levels of HbA1c, FCP, FPG, total cholesterol (TC), TG, HDL-C, and low density lipoprotein-cholesterol (LDL-C) among the four CEA quartiles (all P values for trend < 0.05). There were no significant differences in systolic blood pressure, diastolic blood pressure, ALT, and aspartate aminotransferase (AST).

Table 1

Comparison of clinical characteristics in patients with various CEA levels

Variables	Type 1 diabetes (n = 324)	Type 2 diabetes (n = 3019)	P value
Gender (men)	167 (51.5)	1859 (61.6)	0.001
Current smoker (%)	23 (23.5)	577 (28.0)	0.327
Current drinker (%)	11 (11.2)	268 (13.0)	0.607
Age (years)	55 (45–62)	56 (49–64)	0.012
Duration of diabetes (years)	3 (1–8)	6 (1–10)	< 0.001
BMI (kg/m2)	22.0 (20.1–24.4)	24.5 (22.2–26.8)	< 0.001
Systolic blood pressure (mmHg)	120 (110–135)	130 (120–140)	< 0.001
Diastolic blood pressure (mmHg)	80 (70–83.75)	80 (72–85)	< 0.001
HbA1c (%)	9.5 (7.6–11.6)	8.7 (7.2–10.7)	< 0.001
FPG (mmol/l)	8.24 (6.06–10.99)	7.84 (6.32–10.16)	0.448
FCP (ng/ml)	0.72 (0.22–1.65)	1.71 (1.03–2.45)	< 0.001
TC (mmol/l)	4.70 (4.00–5.32)	4.64 (3.99–5.34)	0.788
TG (mmol/l)	1.02 (0.72–1.54)	1.44 (0.99–2.15)	< 0.001
HDL-C (mmol/l)	1.20 (0.99–1.50)	1.06 (0.90–1.27)	< 0.001
LDL-C (mmol/l)	2.95 (2.37–3.60)	3.00 (2.42–3.65)	0.439
SCr (umol/l)	60 (50–72)	65 (54–76)	< 0.001
ALT (U/l)	17 (12–27)	20 (14–30)	< 0.001
AST (U/l)	18 (15–23)	19 (15–24)	0.276
CEA (ng/ml)	2.79 (1.86–4.02)	2.34 (1.62–3.42)	< 0.001

Data are median (interquartile range, IQR) or number (%). P values between two groups were calculated using Mann–Whitney test for continuous variables or Chi square test for categorical variables

ALT alanine aminotransferase, AST aspartate aminotransferase, BMI body mass index, CEA carcinoembryonic antigen, FPG, fasting plasma glucose, HbA1c glycosylated hemoglobin, TC total cholesterol, TG triglyceride, HDL-C high density lipoprotein-cholesterol, LDL-C low density lipoprotein-cholesterol, FCP fasting C peptide, SCr serum creatinine

Table 2

Comparison of clinical characteristics in patients with various CEA levels

Variables	Q1 (n = 835)	Q2 (n = 827)	Q3 (n = 844)	Q4 (n = 837)	P value
Gender (men)	415 (49.7)	474 (57.3)	544 (64.5)	593 (70.8)	< 0.001
Current smoker (%)	98 (17.6)	126 (23.7)	160 (29.5)	216 (41.1)	< 0.001
Current drinker (%)	57 (10.2)	48 (9)	88 (16.2)	86 (16.4)	< 0.001
Age (years)	54 (46–61)	56 (48–63)	56 (50–64)	58 (51–65)	< 0.001
Duration of diabetes (years)	5 (1–10)	6 (1–10)	5 (1–10)	5 (1–10)	0.013
BMI (kg/m2)	24.6 (22.2–26.8)	24.3 (22.2–26.8)	24.1 (21.8–26.8)	23.9 (21.5–26.2)	0.004
Systolic blood pressure (mmHg)	130 (120–140)	130 (120–140)	130 (120–140)	130 (120–140)	0.329
Diastolic blood pressure (mmHg)	80 (70–85)	80 (72–85)	80 (70–86)	80 (70–85)	0.782
HbA1c (%)	7.8 (6.7–9.2)	8.5 (7.0–10.4)	9.1 (7.5–11.0)	10.1 (8.1–12.0)	< 0.001
FPG (mmol/l)	7.24 (6.00–8.83)	7.68 (6.30–10.16)	8.15 (6.50–10.50)	8.57 (6.68–11.47)	< 0.001
FCP (ng/ml)	1.84 (1.23–2.58)	1.78 (1.10–2.54)	1.57 (0.85–2.37)	1.31 (0.67–2.08)	< 0.001
TC (mmol/l)	4.50 (3.88–5.22)	4.70 (4.00–5.40)	4.65 (4.01–5.37)	4.73 (4.06–5.40)	< 0.001
TG (mmol/l)	1.45 (0.97–2.11)	1.47 (1.05–2.20)	1.37 (0.93–2.12)	1.32 (0.89–1.99)	< 0.001
HDL-C (mmol/l)	1.05 (0.90–1.25)	1.06 (0.90–1.27)	1.08 (0.90–1.29)	1.09 (0.90–1.33)	0.007
LDL-C (mmol/l)	2.96 (2.30–3.52)	3.06 (2.50–3.72)	2.96 (2.40–3.63)	3.00 (2.50–3.71)	0.022
SCr (umol/l)	61 (51–72)	64 (53–74)	65 (55–77)	66 (56–78)	< 0.001
ALT (U/l)	19 (13–30)	19 (14–30)	20 (14–30)	19 (13–30)	0.587
AST (U/l)	18 (15–24)	18 (15–24)	19 (15–24)	19 (15–25)	0.445

Data were median (interquartile range, IQR) or number (%). P values among four groups were calculated using Kruskal–Wallis test for continuous variables or Chi square test for categorical variables

Levels of CEA in Different Subgroups

Also, the levels of CEA in different subgroups were also investigated in our study (Fig. 1). Data indicated that men, current smokers, current drinkers, the elderly, and patients with insulin deficiency had higher CEA levels than their counterparts (P < 0.001). Individuals with type 1 diabetes had higher CEA than those with type 2 diabetes. Moreover, data indicated that levels of CEA were different among patients with different HbA1c intervals grouped by HbA1c < 7%, 7.0–8.9%, and ≥ 9.0% (P value for trend < 0.001). There was no significant difference in CEA levels among patients with different diabetes duration. However, overweight/obesity subjects had lower CEA levels than those with normal weight (P < 0.05).

Fig. 1

Levels of CEA in different subgroups. BMI body mass index, CEA carcinoembryonic antigen, FCP fasting C peptide, HbA1c glycosylated hemoglobin

Clinical Parameters Related to CEA Level

Multiple linear regression analysis indicated that variables such as current smoking, age, BMI, and HbA1c were independently correlated with CEA levels (all P < 0.05, shown in Table 3). Simultaneously, the linear association between age/HbA1c and CEA was fitted, controlling the influence of confounder factors such as drinking status, gender, diabetes duration, FCP, BMI, TG, and HDL-C (Fig. 2). The respective equations of CEA (ng/ml) with age (years) and HbA1c (%) were Y = 0.037X + 0.7, and Y = 0.175X + 1.37 (all P values < 0.001).

Table 3

Clinical indices related to CEA

Variables	B	Standard deviation	t	P value
Current smoker (yes)	0.708	0.352	2.009	0.045
Age (years)	0.038	0.014	2.665	0.008
BMI (kg/m2)	− 0.086	0.043	− 2.004	0.045
HbA1c (%)	0.163	0.068	2.400	0.016
(Constant)	1.325	1.599	0.829	0.407

Data were analyzed using multiple linear regression by backward method, adjusted for drinking status, gender, duration of diabetes, diabetes types, TC, HDL-C, and FCP

Fig. 2

Association of CEA with age (a) and HbA1c (b). The respective linear regression equations were Y = 0.037X + 0.7, and Y = 0.175X + 1.37 (all P values < 0.001), where X is age (years) or HbA1c (%) and Y is CEA level (ng/ml). Data were adjusted for drinking status, gender, diabetes duration, FCP, BMI, TG, and HDL-C

Discussion

In this study, we reported an intimate correlation of CEA with clinical parameters such as current smoking, age, BMI, and HbA1c in patients with diabetes.

Tobacco smoking has been reported to associated with increased cancer incidence and mortality [24]. In this study, data indicated a higher level of CEA in current smokers.

The elderly have a higher incidence of cancer due to impaired immune function [25]. Aging was also reported to be associated with elevated CEA early in 1976 [26], and positive association between age and CEA has also been observed in current study. Our previous study demonstrated that BMI was positively associated with FCP, and overweight/obesity subjects had better glycemic control [27], which may explain the lower levels of CEA in those with overweight/obesity in comparison with normal weight individuals.

Elevated CEA levels in patients with pancreatic cancer have been reported to be associated with onset of diabetes [28]. It has also been reported that patients with diabetes have elevated levels of tumor-related markers, such as CA 19-9, CA125, CA153, in previous studies [9, 18]. It has been demonstrated that glycemic levels were positively associated with tumor markers such as CA 19-9 [18]. In this study, we report a positive relationship between HbA1c and CEA. The mechanism behind this phenomenon was unclear. Chronic inflammation is associated with insulin resistance and the onset of diabetes. It was reported that patients with diabetes had increased inflammatory molecules such as C-reactive protein (CRP), adiponectin, interleukin-6, and adhesion molecules such as soluble E-selectin, ICAM-1, etc. [29-32]. On the other hand, chronic inflammation induced the metabolic reprogramming associated with tumorigenesis of colorectal cancer [33]. It has also been reported that hyperglycemia may modulate the activity of the rate-limiting enzyme, glutamine:fructose-6-phosphate amidotransferase (GFAT), resulting in an increase of hexosamine biosynthetic pathway (HBP) activity and increased cell proliferation, invasion, and tumor progression of colon cancer [34]. These phenomena may partially explain elevated CEA levels in individuals with poor glycemic control.

However, there are some limitations in this study. Firstly, this was a cross-sectional study, it is unknown if this intimate relationship between HbA1c and CEA means a higher risk of carcinoma incidence, and more studies are warranted to investigate whether elevated CEA is associated with increased carcinoma risk in patients with diabetes. Secondly, the mechanism of the influence of HbA1c on CEA needs more experimental studies; this was not investigated here. Finally, subjects with normal glycemic status were not included as a control group in this study.

Conclusion

HbA1c is positively correlated with CEA in patients with diabetes. We should be scrupulous in interpreting the results of CEA in patients with diabetes, especially in those with poor glycemic control.