Assessment of Glucose and HbA1c Monitoring in a Pancreatic Cancer Surveillance Program for High-Risk Individuals

Abstract

INTRODUCTION:

Several studies suggest that new-onset diabetes mellitus is an early manifestation of pancreatic ductal adenocarcinoma (PDAC). Therefore, the International Cancer of the Pancreas Screening Consortium recommends glucose and hemoglobin A1c (HbA1c) monitoring in high-risk individuals (HRIs) undergoing surveillance. However, evidence that such monitoring improves PDAC detection is lacking. Our aim was to investigate the association between serum glucose and HbA1c values and the development of PDAC in HRIs undergoing surveillance.

METHODS:

Participants were recruited from the familial pancreatic cancer surveillance cohort, which follows hereditary predisposed HRIs yearly by magnetic resonance imaging and/or endoscopic ultrasound and blood sampling. Those who underwent fasting glucose and/or HbA1c monitoring at least once were eligible candidates.

RESULTS:

Four hundred four HRIs met the inclusion criteria. During a median follow-up of 41 months (range 14–120), 9 individuals developed PDAC and 4 (without PDAC) were diagnosed with new-onset diabetes mellitus. Glucose levels ranged from 3.4 to 10.7 mmol/L (mean 5.6 ± 0.7) and HbA1c levels from 25 to 68 mmol/mol (mean 37.7 ± 4.1). The mean values did not differ significantly between PDAC cases and controls. The percentage of individuals with at least one elevated value were comparable between PDAC cases and controls for glucose (33% and 27%, P = 0.707) and HbA1c (22% and 14%, P = 0.623). No consistent glucose or HbA1c trends over time suggested a correlation with PDAC development.

DISCUSSION:

In this HRI surveillance cohort, measuring glucose and HbA1c values did not contribute to PDAC detection. Larger and longer-term studies are needed to determine the final role of glucose and HbA1c monitoring in PDAC surveillance.

INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC) is a devastating disease with a 5-year mortality rate exceeding 90% (1). Surgical resection at an early stage provides the only chance for cure (2). However, the time between appearance of a visible lesion on imaging and development of metastasis is short (3). Thus, timely PDAC detection is essential but challenging.

Several studies investigated the association between diabetes mellitus (DM) and PDAC. DM can be seen as an early manifestation of PDAC and as a risk factor (4). Approximately 25% of the patients with PDAC develop new-onset diabetes (NOD), 1–3 years before their cancer diagnosis (5–7). In addition, 0.4%–0.8% of patients aged older than 50 years with NOD will be diagnosed with PDAC within 3 years (8). In addition, patients with recent-onset DM (<1 year) have a higher PDAC risk than those with long-standing DM (relative risk 5.38 vs 1.95) (9). Finally, DM often improves after pancreatic cancer surgery, suggesting cancer being the cause (10,11).

Owing to the relatively low incidence of PDAC (7.7 per 100,000 person-years), population-based screening is unfeasible (1). In addition, screening in patients with NOD seems ineffective and is not recommended (12). Meanwhile, individuals with familial pancreatic cancer (FPC) and/or a known genetic predisposition form a well-defined (study) population for PDAC surveillance. For this group of high-risk individuals (HRIs), experts from the international Cancer of the Pancreas Screening Consortium reached consensus that NOD should prompt further investigation for the presence of pancreatic cancer. Therefore, they recommend routine testing of fasting blood glucose and/or hemoglobin A1c (HbA1c) (8).

So far, there is no direct evidence that glucose and HbA1c monitoring improves the detection of PDAC. The aim of this study was to assess its association in a PDAC surveillance program for individuals at increased hereditary risk.

METHODS

Study design and surveillance protocol

This study is part of the FPC surveillance study (NL40489.078.13), a multicenter prospective cohort study conducted in 3 academic centers in the Netherlands (Erasmus Medical Center Rotterdam, Amsterdam University Medical Center and University Medical Center Utrecht) since 2006. After genetic evaluation, individuals with a family history of PDAC (2 or more first-degree relatives) and/or a germline mutation with an estimated 10-fold increased risk of developing PDAC are being followed yearly by magnetic resonance imaging and/or endoscopic ultrasound and blood sampling. Complete inclusion and exclusion criteria are supplied in the online Supplementary Digital Content (see Supplementary File, http://links.lww.com/CTG/B208) (3). Fasting blood glucose and HbA1c levels have been prospectively collected per protocol at each follow-up visit since 2008. Objectives of this study were (i) to compare the number of individuals who developed NOD between the PDAC cases and controls without PDAC, and (ii) to determine the range and variability of glucose and HbA1c levels in this surveillance population, and to compare these values and the course of the levels over time between PDAC cases and controls.

Study population and data collection

Individuals with at least one fasting blood glucose or HbA1c measurement were included in the present analysis. Participants using glucose lowering drugs or steroids were excluded. In addition, values determined during the temporary use of these drugs were excluded from the analysis. Baseline characteristics, including age, gender, race, smoking, alcohol use, body mass index, and risk category (family history of PDAC or type of genetic mutation) were derived from the study database. Missing data were supplemented from the electronic medical records. HbA1c values in percentages were converted to mmol/mol according to the formula (10.93 × %) − 23.5 (13). Glucose levels can be converted from mmol/L to mg/dL by multiplying by 18. HbA1c levels (in mmol/mol) do not require conversion. The cutoff for fasting glucose was set at 6.1 mmol/L and for HbA1c at 43 mmol/mol, according to local standards (14).

Data and statistical analysis

Descriptive data were reported as means with corresponding SD or medians with interquartile ranges. The last glucose or HbA1c value determined during the follow-up was used to calculate the mean value. Differences in continuous variables with normal distribution were assessed with an unpaired t test. For categorical variables, frequencies were noted in percentages, and a Pearson χ² test was used to compare groups. Two-sided P values ≤0.05 were considered statistically significant. Data were analyzed by IBM SPSS Statistics 15.0 for Windows (SPSS, Chicago, IL). R Core Team (2020) was used for the generation of graphs.

The necessary ethical approval/clearance was obtained.

RESULTS

Study population

In October 2021, 463 HRIs participated in the FPC study. 20 individuals with preexisting DM and 2 who used steroids were excluded. In 37 individuals, no glucose or HbA1c values had been assessed. The remaining 404 individuals were included in the present analysis. Table 1 presents their baseline characteristics.

Table 1.

Baseline characteristics

Variables	Non-PDAC (n = 395)	PDAC (n = 9)	P value
Age, mean (SD), yr	54 (9.6)	54 (10.3)	0.772
Male, n (%)	161 (41)	6 (67)	0.123
Risk category, n (%)			<0.001
Member of FPC family	221 (56)	0
Mutation carrier	174 (44)	9 (100)
Peutz-Jeghers syndrome	9 (5)	2 (22)
BRCA1 (HBOC syndrome)	6 (4)	—
BRCA2 (HBOC syndrome)	42 (24)	—
PALB2 (HBOC syndrome)	2 (1)	—
CDKN2A (FAMMM syndrome)	106 (61)	7 (78)
MLH1/MSH2/MSH6 (Lynch syndrome)	2 (1)	—
TP53 (Li-Fraumeni syndrome)	4 (2)	—
Other	3 (2)	—
BMI, kg/m2, n (%)			0.256
Underweight (≤18.4)	7 (2)	0
Normal (18.5–24.9)	177 (45)	4 (45)
Overweight (25.0–29.9)	162 (41)	2 (22)
Obese (≥30.0)	49 (12)	3 (33)

BMI, body mass index; BRCA1, breast cancer gene 1; CDKN2A, cyclin dependent kinase inhibitor 2A; FAMMM, familial atypical multiple mole melanoma; FPC, familial pancreatic cancer; HBOC, hereditary breast and ovarian cancer; MLH1, MutL homolog 1; MSH2, MutS homolog 2; MSH6, MutS homolog 6; PALB, partner and localizer of BRCA2; PDAC, pancreatic ductal adenocarcinoma; TP53, tumor protein 53.

PDAC and new-onset DM diagnosis

During a median follow-up of 41 months (range 14–120), 9 (2.2%) individuals developed PDAC, all carriers of a known genetic mutation. Four individuals developed NOD requiring glucose-lowering drugs, none of which developed PDAC. Two of these individuals had a follow-up of longer than 3 years since the onset of NOD. Furthermore, 5 individuals were given lifestyle recommendations to control their glucose levels, one of which developed PDAC. Table 2 presents the characteristics of the patients with PDAC.

Table 2.

Characteristics of patients with PDAC

N	Agea	Gender	Risk category	Surgery	Glucose ≥6.1 mmol/Lb	HbA1c ≥ 43 mmol/molb	NOD
1	54	Female	PJS	Yes	No	No	No
2	65	Female	PJS	Yesc	No	No	No
3	42	Male	CDKN2A	Yes	Yes	No	No
4	68	Male	CDKN2A	Yes	Yes	No	Nod
5	68	Male	CDKN2A	No	Yes	Yes	No
6	55	Female	CDKN2A	No	No	Yes	No
7	55	Male	CDKN2A	Yes	No	No	No
8	50	Male	CDKN2A	Yes	No	No	No
9	53	Male	CDKN2A	Yes	No	No	No

CDKN2A, cyclin dependent kinase inhibitor 2A; NOD, new-onset diabetes mellitus; PDAC, pancreatic ductal adenocarcinoma; PJS, Peutz-Jeghers syndrome.

^aAt diagnosis.

^bAt least once.

^cSurgery discontinued due to advanced disease.

^dLifestyle interventions to control glucose, however not officially diagnosed by the general practitioner.

Glucose and HbA1c outcomes

Glucose levels ranged from 3.4 to 10.7 mmol/L (mean 5.6 ± 0.7) (Table 3) and were elevated at least once in 109 individuals (27%). The mean values did not differ between PDAC cases (5.8 ± 1.3) and controls (5.6 ± 0.7, P = 0.252), nor did the percentage of individuals with at least one elevated value (33% cases vs 27% controls, P = 0.707). In addition, the mean percentage changes from baseline to the last follow-up visit did not differ between PDAC cases (8.3% ± 17.0%) and controls (4.5% ± 12.1%; P = 0.179) (Table 4). Similarly, the percentage changes from 3 to 1 year before PDAC diagnosis or the last visit did not differ (7.4% ± 17.1% cases vs 3.0% ± 11.8% controls; P = 0.137) (Table 5).

Table 3.

Glucose and HbA1c values

	Non-PDAC (n = 395)	PDAC (n = 9)	Total cohort (n = 404)	P value
Overview of glucose values (mmol/L)
Range	3.4–10.7	3.8–7.8	3.4–10.7	—
Mean (SD)	5.6 (0.7)	5.8 (1.3)	5.6 (0.7)	0.252
Glucose ≥6.1	N = 106 (27%)	N = 3 (33%)	N = 109 (27%)	0.707
Overview of HbA1c values (mmol/mol)
Range	25.0–68.0	31.0–51.0	25.0–68.0	—
Mean (SD)	37.7 (4.1)	38.9 (5.6)	37.7 (4.1)	0.192
HbA1c ≥ 43	N = 56 (14%)	N = 2 (22%)	N = 58 (14%)	0.623

PDAC, pancreatic ductal adenocarcinoma.

Table 4.

Changes in glucose and HbA1c values from baseline to the last follow-up visit

	Non-PDAC (n = 343)	PDAC (n = 9)	Total cohort (n = 352)	P value
Glucose (%)
Range	−33.3 to 57.4	−11.6 to 44.4	−33.3 to 57.4	—
Mean (SD)	4.5 (12.1)	8.3 (17.0)	4.6 (12.2)	0.179
HbA1c (%)
Range	−35.3 to 57.9	−16.2 to 30.8	−35.3 to 57.9	—
Mean (SD)	2.4 (7.8)	5.1 (14.5)	2.5 (8.0)	0.305

PDAC, pancreatic ductal adenocarcinoma.

Table 5.

Changes in glucose values from 3 to 1 year before PDAC diagnosis or the last follow-up for controls to the last follow-up visit (%)

	Non-PDAC (n = 338)	PDAC (n = 9)	Total cohort (n = 347)	P value
Range	−33.3 to 78.7	−11.6 to 44.4	−33.3 to 78.7	—
Mean (SD)	3.0 (11.8)	7.4 (17.1)	3.1 (11.9)	0.137

PDAC, pancreatic ductal adenocarcinoma.

HbA1c levels varied between 25 and 68 mmol/mol (mean 37.7 ± 4.1) and were elevated at least once in 58 individuals (14%) (Table 3). The mean values did not differ between PDAC cases (38.9 ± 5.6) and controls (37.7 ± 4.1, P = 0.192), neither did the percentages of individuals with elevated values (22% cases and 14% controls, P = 0.623). Furthermore, the mean percentage changes from baseline to the final visit were similar for PDAC cases (5.1% ± 14.5%) and controls (2.4% ± 7.8%; P = 0.305) (Table 4). Similarly, the percentage changes from 3 to 1 year before PDAC diagnosis or the last visit did not differ between cases (4.1% ± 14.7%) and controls (2.2% ± 6.9%; P = 0.361) (Table 6).

Table 6.

Changes in HbA1c values from 3 to 1 years before PDAC diagnosis or last follow-up for controls to last follow-up visit (%)

	Non-PDAC (n = 336)	PDAC (n = 8)	Total cohort (n = 344)	P value
Range	−19.05 to 50.0	−16.2 to 30.8	−19.1 to 50.0	—
Mean (SD)	2.2 (6.9)	4.1 (14.7)	2.2 (7.2)	0.361

PDAC, pancreatic ductal adenocarcinoma.

Figure 1 demonstrates the individual courses of glucose (mmol/L) and HbA1c (mmol/mol) levels over time, until the moment of PDAC diagnosis or last surveillance visit for controls. No gradual increase in glucose or HbA1c levels was observed before PDAC detection.

Figure 1.

Glucose and HbA1c levels over time of PDAC cases and controls. PDAC, pancreatic ductal adenocarcinoma.

DISCUSSION

This study evaluates the role of hyperglycemia as a predictive marker for PDAC development in a surveillance program for individuals at increased heredity risk. Our results show that NOD was not more common in PDAC cases. Furthermore, glucose and HbA1c levels were highly variable and not predictive for PDAC. In addition, individual changes in these levels over time did not show an association with PDAC. However, it is important to note that the limited number of PDAC cases prohibits drawing definitive conclusions.

A possible explanation for the lack of association between NOD and PDAC may be that the used cutoff values are too high, as they are designed for detecting diabetes to prevent associated cardiovascular complications, rather than to detect subtle elevations predictive for PDAC. In addition, it may be difficult to distinguish NOD due to PDAC from the more common type 2 DM. Using a clinical prediction model, such as The Enriching New-Onset Diabetes for Pancreatic Cancer model which has recently been validated, can help to determine the risk of pancreatic cancer in patients with NOD in the future (15–17).

Obviously, glucose levels are easily influenced by external factors, such as stress and illness, which may cause fluctuating values (6,18). In addition, glycemic values are influenced by weight. While weight gain may increase values, weight loss in combination with high glucose levels may be suggestive for PDAC (18). HbA1c does not undergo daily fluctuations, which may make it a more reliable marker. However, we did not find HbA1c to be predictive of PDAC development in our population.

A retrospective case-control study did find an association between fasting blood glucose levels and time to PDAC diagnosis, tumor volume, and grade (19). Glucose levels increased gradually until diagnosis of PDAC, and hyperglycemia was associated with pancreatic tumor volumes of >1.1 mL and invasive cancer, providing an opportunity for early diagnosis of PDAC. A second retrospective study showed that HRIs undergoing pancreatic cancer surveillance with elevated HbA1c values (≥40 mmol/mol) were more likely to have pancreatic cysts compared with HRIs with normal values (20). How to interpret this finding is difficult. Whether the presence of a cystic neoplasm or the malignant progression of such a lesion influences blood sugar levels will need to be further investigated, in addition to whether the presence of such a lesion in an HRI increases their risk of PDAC (21).

Apart from our study, one other prospective study sought to determine if NOD may serve as a marker for early PDAC (22). It contained 100 HRIs who underwent one-time screening by endoscopic ultrasound and fasting blood glucose measurement. They reported one patient with NOD (diagnosed by a fasting glucose ≥7 mmol/L), which was also the single patient who was found to have PDAC. The authors assert that their results support the association between NOD, a conclusion that our results do not endorse.

To date, this is the only other study that assessed NOD, glucose, and HbA1c monitoring in a PDAC surveillance program for HRIs. It is one of the largest prospectively followed high-risk cohorts, with a stringent protocol of yearly surveillance, allowing for multiple measurements over time. A major limitation is the low number of PDAC cases, which prohibits us from establishing definite correlations. However, the aim was primarily to describe our cohort. Furthermore, glucose and HbA1c were not always monitored in the past. In addition, the fact that all PDAC cases occurred in patients with Peutz-Jeghers syndrome and familial atypical multiple mole melanoma suggests that the findings may not be fully generalizable to other hereditary conditions.

In conclusion, our data did not show an association between PDAC and NOD in a surveillance population of hereditary predisposed HRIs. Elevated glucose and/or HbA1c values were frequent phenomena but did not correlate with PDAC presence or development. Further research is needed to determine the true value of glycemic monitoring for early PDAC detection. Our Cancer of the Pancreas Screening surveillance cohort will likely offer more conclusive answers in the future, when more PDAC cases have developed and more data have been assembled.

CONFLICTS OF INTEREST

Guarantor of the article: Djuna L. Cahen, MD, PhD.

Specific author contributions: J.M. conceptualization, data collection, analysis and writing–original draft. J.G.Y.d.J.: data collection, analysis and writing–concept draft. G.M.F.: supervisor, writing–review & editing. B.D.M.K.: data collection. I.J.M.L.: data collection. P.F.: writing–review & editing. F.P.V.: writing–review & editing. M.J.B.: supervisor, writing–review & editing. D.L.C.: supervisor, writing–review & editing. The authors approved the final draft.

Financial support: None to report.

Potential competing interests: None to report.

Study Highlights.

WHAT IS KNOWN

• ✓ New-onset diabetes may be an early sign of pancreatic ductal adenocarcinoma (PDAC).

• ✓ As such, guidelines advocate glucose and HbA1c monitoring for high-risk individuals under PDAC surveillance.

• ✓ Evidence that glucose and HbA1c monitoring improves PDAC detection is lacking.

WHAT IS NEW HERE

• ✓ In a large cohort of hereditary predisposed individuals, new-onset diabetes was not more common in patients with PDAC.

• ✓ In addition, glucose and HbA1c levels were not predictive for the development of PDAC.

• ✓ In conclusion, glucose and HbA1c monitoring did not contribute to the detection of PDAC in a surveillance program for individuals at heredity increased risk of pancreatic cancer.

Supplementary Material

ct9-15-e00777-s001.docx ^{(12.6KB, docx)}