Association of Type O Blood with Pancreatic Neuroendocrine Tumors in Von Hippel–Lindau Syndrome

Abstract

Background

ABO blood type antigens are expressed not only on human red blood cells, but also throughout the gastrointestinal tract and in normal pancreatic tissue. Previous studies have identified an association between ABO blood type and various malignancies. We analyzed the association of ABO blood type with pancreatic neuroendocrine tumors (PNETs) in a high-risk cohort of patients with Von Hippel–Lindau (VHL) syndrome.

Methods

A retrospective review was performed of 798 patients with VHL syndrome. Blood type was confirmed for 181 patients. Fisher’s exact test and Mehta’s modification to Fisher’s exact test were used to test for an association between ABO blood type and manifestations of VHL syndrome.

Results

We found a strong trend for association between O blood type and pancreatic disease manifestation in patients with VHL syndrome (P = 0.047). More importantly, there was a significant association of O blood type with solid pancreatic lesions consistent with PNETs (P = 0.0084). Patients with solid pancreatic lesions who met criteria for surgical resection at the National Institutes of Health also had a higher rate of O blood type than those who did not require surgery (P = 0.051).

Conclusions

Our findings suggest an association between O blood type and pancreatic manifestation of disease in patients with VHL syndrome, especially for PNETs. Screening and surveillance approaches for pancreatic lesions in patients with VHL syndrome should also consider patient blood type. The possibility of A, B, H misexpression in PNETs should also be explored to determine whether the serologic association with disease translates into a relationship with tissue pathology.

ABO blood type system is defined by the expression of human blood group antigens on red blood cells: type A, B, AB, or O. However, the glycoproteins that form these antigens are not limited to expression on erythrocytes. Similar blood group antigen expression patterns have been found in normal tissue throughout the body, including in the gastrointestinal tract and pancreas.¹

For several decades, numerous studies have suggested an association between ABO blood type and various malignancies.¹ For example, several investigators have identified an association between blood group A and increased risk of developing pancreatic adenocarcinoma.^1–6 Furthermore, histologic analysis has shown alterations in blood group antigen expression in patient derived pancreatic tumor cells.^7,8 Although multiple investigators have suggested an association between specific blood group and exocrine pancreatic tumors, it has not been investigated in pancreatic neuroendocrine tumors (PNETS). We are also not aware of any studies evaluating ABO blood type and risk of tumor manifestations in familial cancer syndromes, a high-risk population. Given that PNETS may be associated with familial cancer syndromes such as Von Hippel–Lindau (VHL) and multiple endocrine neoplasia type 1, determining the presence of an association between blood type and pancreatic neuroendocrine manifestation may have important implications for surveillance and screening for PNETS in such families because of the risk that these lesions may progress or be metastatic.^9,10

Patients with VHL syndrome have a high risk of developing multiple neoplasms throughout the body. Patients with VHL syndrome have a 8–17% risk for developing both benign and malignant PNETS.¹¹ In this study, we analyzed the association of ABO blood type and pancreatic manifestations in patients with VHL syndrome.

MATERIALS AND METHODS

Patients

The medical records of 798 patients with VHL syndrome who were enrolled onto prospective protocols approved by our institutional review board at the National Institutes of Health (NIH) Clinical Center were reviewed. Of these, 181 patients had their blood type recorded in their medical record with laboratory confirmation. Further information was collected for these 181 patients, including demographics (age, sex, and race), manifestations of their VHL syndrome, and genotype (type of mutation and exon mutation).

Data Analyses

Patient blood types were evaluated in three ways: the first level included blood type and Rh factor for eight categories (A+, A−, AB+, B+, B−, O+, and O−); the second level looked at blood type (A, AB, B, and O); and the last evaluated type O versus all non-O blood types.

The presence or absence of disease in six organs (central nervous system, ear, eye, kidney, adrenal gland and pancreas) was recorded (Fig. 1). This information was collected from medical records, radiologic studies, and/or pathology. Pancreatic disease was categorized according to presence of cystic disease and/or solid pancreatic lesions on computed tomography (CT) and magnetic resonance imaging (MRI). Solid and complex pancreatic lesions which enhanced with contrast on CT and or MRI where considered consistent with PNETS in addition to other imaging features.^12,13 These manifestations were analyzed independently for a relationship with blood type; however, 77 patients had both cystic and solid lesions on imaging. Patients with solid pancreatic lesions were stratified to separate those who required surgical intervention and those who had not met criteria for surgery. The criteria used for surgical intervention at our institution is based on tumor size (>2 cm in the head of the pancreas, or >3 cm in other pancreatic locations), tumor growth, and/or suspected presence of metastatic disease.¹⁴ Finally, patients who underwent surgical intervention were stratified according to the presence of metastatic disease.

FIG. 1.

VHL disease manifestations in study cohort. The categories tested for association with blood type are not mutually exclusive as some patients will have multiple manifestations. Thus, subcategories may not add up to 100% or the total number of the category heading because of missing data for some cases

To test for a general association between blood type and disease state, Mehta’s modification to Fisher’s exact test was used for analyses of 8 × 2 and 4 × 2 tables, to analyze blood groups with and without inclusion of Rh factor, respectively. Fisher’s exact test was used for 2 × 2 tables to assess between patients with O blood type and those with non-O blood types. Ages were compared among blood types by a Kruskal–Wallis test, while Mehta’s version of Fisher’s exact test was used to determine the significance of the association of blood type with sex, race and specific exon mutation. As a result of the multiple associations tested in this study, there was concern that each comparison may not be independent. Therefore, a stricter P value threshold was designated to protect against statistical error due to multiple comparisons. P values of <0.01 suggest significant associations, while those for which 0.01 < P <0.05 indicate strong trends. To account for confounding variables, demographic and clinical characteristics (age, sex, race, and genotype) were also analyzed.

RESULTS

ABO blood group types and disease manifestations were recorded for 181 patients with VHL syndrome (Fig. 1). The frequency distribution of ABO blood groups in our cohort was similar to the blood group distribution in the United States (Table 1). There was no significant difference in age, sex, race or mutation status by blood types (Table 2).

TABLE 1.

Comparison of blood types in the United States and blood types identified in our cohort of patients with VHL syndrome

Population	n	Type A+ (%)	Type A− (%)	Type AB+ (%)	Type AB− (%)	Type B+ (%)	Type B− (%)	Type O+ (%)	Type O− (%)	P
United Statesa	307,212,123	35.70	6.30	3.40	0.60	8.50	1.50	37.40	6.60	0.7414b
NIH VHL syndrome cohort	181	63 (34.81)	7 (3.87)	4 (2.21)	1 (0.55)	14 (7.73)	3 (1.66)	78 (43.09)	11 (6.08)

^aFrom “Blood Types in the US”, Bloodcenter.stanford.edu, http://bloodcenter.standford.edu/about_blood/blood_types.html

^bCalculated by χ² test with Monte Carlo estimate for the exact test

TABLE 2.

Clinical characteristics of patients organized by blood type

Characteristic	Total	A	AB	B	O
n	181	70	5	17	89
Age, mean ± SEMa	47.5 ± 0.939b	47.3 ± 1.43	43.2 ± 6.29	52.1 ± 3.38c	47.0 ± 1.37
Sex, M/Fa	86/95	34/36	2/3	8/9	42/47
Racea
Asian	3	2	0	1	0
African American	7	1	0	1	5
Hispanic	1	1	0	0	0
Unknown	9	4	0	0	5
White	161	62	5	15	79
Exona
1	41	15	3	2	21
2	13	8	0	0	5
3	49	20	1	2	26

^aNo significant difference for each blood type from total group in patients who underwent genetic testing

^bn = 180

^cn = 16

Differences in blood group expression in six organ systems involved in VHL syndrome were analyzed (Tables 3, 4). Results are not presented for analyses of blood type and Rh factor because no potentially important associations were detected. A strong trend was seen in the distribution of O blood type between patients with and without pancreatic manifestations of VHL syndrome (52.3% with pancreatic manifestations have O blood type vs. 29.2% of those with nonpancreatic manifestations; P = 0.047), suggesting that patients with pancreatic disease were more likely to have type O blood.

TABLE 3.

Distribution of blood type (non-O types vs. O type) in study cohort stratified by manifestation of VHL syndrome

Disease manifestation	Disease present (n)	Non-O blood type (%)	O blood type (%)	P
Pancreas	Yes (155)	47.7	52.26	0.047
	No (24)	70.8	29.17
Cystic disease	Yes (123)	52.9	47.15	0.046
	No (32)	31.3	68.75
Solid tumors	Yes (111)	41.4	58.56	0.0084
	No (46)	65.2	34.78
Pancreatic surgery	Yes (47)	29.8	70.21	0.051
	No (64)	50.0	50.00
Central nervous system	Yes (140)	53.1	47.86	0.72
	No (39)	48.7	51.28
Eye	Yes (46)	63.0	36.96	0.059
	No (130)	46.2	53.85
Ear	Yes (27)	55.6	44.44	0.68
	No (149)	49.7	50.34
Kidney	Yes (151)	50.3	49.67	1.0
	No (27)	51.9	48.15
Adrenal gland	Yes (45)	62.2	37.78	0.086
	No (132)	47.0	53.03
Normal distribution in the United States		56.0	44.0

TABLE 4.

Distribution of blood type (ABO) in study cohort, stratified by manifestation of VHL syndrome

Disease manifestation	Disease present (n)	Type A (%)	Type AB (%)	Type B (%)	Type O (%)	P
Pancreas	Yes (155)	36.8	2.58	8.39	52.3	0.097
	No (24)	50.0	4.17	16.7	29.2
Cystic disease	Yes (123)	41.5	2.44	8.94	47.1	0.13
	No (32)	21.9	3.13	6.25	68.7
Solid tumors	Yes (111)	33.3	0.90	7.21	58.6	0.016
	No (46)	47.8	6.52	10.8	34.8
Pancreatic surgery	Yes (47)	25.5	0.00	4.26	70.2	0.14
	No (64)	39.1	1.56	9.38	50.0
Central nervous system	Yes (140)	39.3	2.86	10.0	47.9	0.97
	No (39)	38.5	2.56	7.69	51.3
Eye	Yes (46)	50.0	4.35	8.70	37.0	0.14
	No (130)	33.9	2.31	10.0	53.9
Ear	Yes (27)	48.1	0.00	7.41	44.4	0.74
	No (149)	36.2	3.36	10.1	50.3
Kidney	Yes (151)	39.1	3.31	7.95	49.7	0.35
	No (27)	33.3	0.00	18.5	48.1
Adrenal gland	Yes (45)	53.3	0.00	8.89	37.8	0.086
	No (132)	33.3	3.79	9.87	53.0
Normal distribution in the United States		42.0	4.00	10.0	44.0

Patients with VHL syndrome and pancreatic disease were separated into those with cystic and solid manifestations on the basis of a pancreatic protocol CT scan and MRI with solid or complex lesions showing arterial enhancement. Among patients with cystic disease, 47.2% have O blood type, while 68.7% of patients without evidence of cystic disease have O blood type P < 0.05). Patients with solid tumors were significantly more likely to have O blood type (58.6%) compared to those with cystic manifestations of disease (34.8%, P = 0.0084). When compared amongst individual blood groups (A/B/AB/O), there was a strong trend toward an association between blood type and patients with solid tumors vs. those without solid tumors (P = 0.016).

Patients with solid tumors on imaging were further divided into a group who required surgical intervention and those who did not. O blood type was present in 70.2% of those undergoing pancreatic surgery, while 50% of those not undergoing surgery had O blood type. (P = 0.051). Pathology reports were available for 27 of the 47 surgical patients. Of these, 26 patients were diagnosed with a pancreatic neuroendocrine tumor, while one patient had a microcystic serous cystadenoma. We did not find an association between blood type and the presence of metastatic disease in those patients who underwent surgical resection of their pancreatic neuroendocrine tumor.

DISCUSSION

Our results suggest that there is an association between type O blood and pancreatic manifestation in patients with VHL syndrome. The association was strongest for the presence of a solid pancreatic lesion on CT scan (most consistent with neuroendocrine tumor) and O blood type (P = 0.0084). There was also a trend between patients with O blood type and those who met the criteria for surgical intervention at the NIH for a solid pancreatic lesion consistent with a pancreatic neuroendocrine tumor (P = 0.05).

Several hypotheses have been put forth to account for the association of blood type and cancer. One theory is that the ABO gene products have a direct role and function in tumorigenesis.^1,7 Several studies have reported altered antigen expression in neoplastic tissue, as well as an association between altered antigen expression and metastasis/prognosis, showing more aggressive disease and decreased survival rates with decreased tissue ABO expression..^{7,8, 15–19} This suggests that the ABO glycosyltransferase or the glycosylation pattern itself may play a role. It has also been proposed that alterations in tissue inflammatory state or intercellular adhesion may also contribute to the association between blood type and cancer development or risk, as two genome-wide association studies have shown significant associations between the ABO locus and serum levels of both tumor necrosis factor-alpha and intercellular adhesion molecule 1.^20,21 Another possible mechanism for the association between ABO blood type and cancer is the secretory phenotype (proposed to be a protective mechanism) which is more common in O blood type.^6,22 Also, single nucleotide polymorphism and or mutations in fucosyltransferases and glycosyltransferases have been associated with a higher risk of cancer in some but not all studies.^6,23–26

Our findings of an association between O blood type and pancreatic manifestation of VHL syndrome, pancreatic neuroendocrine tumor and a trend in requiring surgical intervention has important clinical ramifications in the surveillance and or screening of patients with VHL syndrome for pancreatic manifestation. Currently, there are few data on what are the most effective approach for screening and surveillance for pancreatic manifestation of VHL syndrome. In our clinical research protocol, we perform follow-up axial imaging (CT and MRI) annually in patients with solid and complex lesions and every 2 years in those with only cystic pancreatic lesions if they remain asymptomatic. Our results would suggest that this approach may be more appropriate for individuals with O blood type and those with other blood type may be followed less frequently depending on the size and location of their solid pancreatic tumors. Of course, because there is significant overlap between these two groups, longer follow-up of these patients will be needed to better define what approach for surveillance is most appropriate.

Our study has several limitations, including a selection bias due to the retrospective nature of this analysis and because confirmed blood type information was available only in a subset of the entire VHL syndrome cohort evaluated at our institution. One of our inclusion criteria was a documented blood type; however, blood type information was often available as a result of presurgical laboratory tests, thus likely overrepresenting individuals that had neurologic, urologic and/or pancreatic manifestations of VHL syndrome as many of these lesions may require surgical intervention. Other factors for selection bias include the over-representation of pancreatic disease (87% vs. the ≤70% reported in the literature).¹¹ We also performed multiple comparisons in the study cohort and false positive associations may be possible. However, to minimize this, we used a lower P-value threshold with P values of <0.01 suggesting a significant associations and P values between 0.01 and <0.05 indicating a statistical trend.

In conclusion, the results of our study suggest an association between blood type-O and PNETS. Further investigation is needed to confirm this finding and describe its applicability to PNETS in sporadic and hereditary settings. Finally, the possibility of A, B, H misexpression in PNETS should be explored to determine if the serologic association with disease translates into a relationship with tissue pathology. Screening and surveillance approaches for pancreatic lesions in patients with VHL syndrome should also consider patient blood type.