ABO Blood Group and Risk of Pancreatic Carcinogenesis in Intraductal Papillary Mucinous Neoplasms

ABO blood group is an inherent factor of individuals and has been implicated in risks of various malignancies including pancreatic cancer (1, 2). Epidemiologic research suggests that individuals with non-O blood group may be at higher risk of pancreatic cancer compared with those with blood group O (3, 4). Genetic variants in the ABO locus at chromosome 9q34.2 have been associated with the risk of this malignancy (5–7). On the basis of these lines of evidence, ABO blood group has been incorporated into prediction models for risk of pancreatic cancer in the general population (8).

Pancreatic cancer currently represents a leading cause of cancer-related death worldwide (9). Despite increasing efficacy of chemotherapeutic regimens (10–12), pancreatic ductal adenocarcinoma (PDAC) has been associated with poor prognosis of patients (9), and therefore, there is a great need for identification of risk factors for this malignancy to implement strategies of early diagnosis and to improve clinical outcomes. Intraductal papillary mucinous neoplasm (IPMN) of the pancreas has been recognized as a precursor lesion of pancreatic carcinoma with variable rates of malignant transformation (13–16). Given relatively slow progression of the IPMNs to invasive carcinoma, a vast majority of patients diagnosed with IPMNs are referred to surveillance programs (17), and surgical resection is recommended when the dilated main pancreatic duct (MPD) or features suggesting development of high-grade dysplasia or carcinoma are observed during the follow-up (18–20). Cohort studies point to cyst characteristics associated with elevated risk of pancreatic carcinogenesis, such as the dilated MPD and large cyst size (13, 14, 21–23). However, host risk factors including genetic and epidemiologic factors remain to be elucidated for this patient population. The association of ABO blood group with pancreatic cancer risk has not been examined in the context of pancreatic carcinogenesis during follow-up of patients with IPMNs.

To test our hypothesis that ABO blood group might be associated with risk of pancreatic cancer among patients with IPMNs, we utilized a large cohort of consecutive patients who were diagnosed with IPMNs during a period of over 25 years. In addition, we evaluated heterogeneity in associations of ABO blood group with pancreatic cancer by cancer types (IPMN-derived carcinoma or concomitant PDAC).

We collected data on consecutive patients diagnosed with IPMNs from our prospectively maintained database (24, 25). Branch-duct and main-duct IPMNs were radiologically defined according to the current consensus guidelines proposed by the International Association of Pancreatology in 2017 (19). We reevaluated the patients who had been diagnosed with IPMNs before the publication of the current version of the guidelines. Diagnosis of branch-duct IPMN was made based on imaging findings of unilocular or multilocular pancreatic cystic lesions that communicate with the MPD. Main-duct IPMN was defined as a lesion with the MPD diameter of >5 mm. To exclude the possibility of preclinical malignancy, we included patients who survived more than six months from the diagnosis of IPMN without a diagnosis of pancreatic carcinoma. Patients were excluded (i) if they had a history of pancreatic carcinoma, (ii) if the IPMN was surgically resected within six months of the diagnosis, (iii) if they were diagnosed with pancreatic carcinoma within six months, or (iv) if the follow-up time was <6 months. This study was conducted according to the guidelines in the Helsinki Declaration and was approved by the ethics committee of the University of Tokyo (Tokyo, Japan; #1804, 2058, and G0500). Informed consent was obtained from the participants on an opt-out basis.

Within our cohort of patients who had been diagnosed with IPMNs at the University of Tokyo Hospital (Tokyo, Japan) between January 1994 and September 2019, we identified 1,815 patients with available data on ABO blood group (Fig. 1). The patients were followed until development of pancreatic carcinoma, death, or the end of follow-up (April 30, 2020), whichever came first. The patients were followed on an outpatient basis every 6–12 months. At every visit, the patients underwent imaging tests (MRI/cholangiopancreatography, contrast-enhanced computed tomography, endoscopic ultrasound, or abdominal ultrasound) along with physical examinations and blood tests including levels of amylase, glucose, CEACAM5 [carcinoembryonic antigen (CEA)], and carbohydrate antigen 19–9 (CA19–9). When a malignant pancreatic lesion was suspected on follow-up imaging studies, we additionally performed endoscopic ultrasound fine-needle aspiration (EUS-FNA) for lesions invading the pancreatic parenchyma and endoscopic retrograde cholangiopancreatography (ERCP) for lesions localizing within the IPMN and/or MPD (26, 27). According to the consensus between Japanese clinicians (19), we did not perform EUS-FNA for cytology and molecular analysis of cyst fluid (28, 29). According to the consensus, which was later published as the international consensus guidelines in 2006 (30), we recommended surgical resection when we observed (i) a mural nodule of the IPMN, (ii) the diameter of the MPD ≥ 10 mm, (iii) positive cytology, or (iv) the diameter of branch-duct IPMN ≥ 30 mm. According to the revision of the guidelines in 2012 (31), we modified our recommendations for surgical resection: that is, a large size of branch-duct IPMN (≥30 mm) was not considered as an absolute indication for surgery. Throughout the study period, we recommended pancreatectomy for symptomatic patients (e.g., patients developing acute pancreatitis without any causes other than IPMN).

We identified 1,815 patients with IPMN with available data on ABO blood group. We collected data on serologically confirmed ABO blood group (A, B, AB, or O) through reviewing electronic medical charts at our institution. For 255 cases (14%) without available data in the system, we additionally utilized self-reported data on ABO blood group (A, B, AB, O, or unknown). Among 418 cases with available data on both serologic and self-reported ABO blood groups, the agreement between these measurements was almost perfect [κ, 0.99; 95% confidence interval (CI), 0.98–1.00]. In our cohort of patients with IPMN, there were no substantial differences in clinical characteristics by the availability of ABO blood group data (Supplementary Table S1). We also collected data on Rh factor (positive or negative); however, a small number of Rh-negative patients in our study population [13 (0.8%) of 1,562 patients with available data] precluded a robust statistical assessment.

When follow-up imaging studies suggested development of pancreatic malignancy, we carried out EUS-FNA and/or ERCP to confirm a pathologic diagnosis (26, 27). For patients who underwent pancreatic tumor resection, the final diagnosis was made based on pathologic examinations of the surgical tissue specimens. When tissue specimens of the tumors were not available, we made a diagnosis of pancreatic carcinoma when typical radiological findings were observed with compatible clinical course. IPMN-derived carcinoma was defined as IPMN with high-grade dysplasia (formerly referred to as carcinoma in situ) or an invasive carcinoma component (32). We assessed the continuity of carcinoma and IPMN during radiological and pathologic examinations to differentiate IPMN-derived carcinoma and concomitant PDAC. Namely, when the continuity (or pathologic transition) between carcinoma and IPMN lesions was observed, the carcinoma was classified as IPMN-derived carcinoma and otherwise, classified as concomitant PDAC.

To supplement the distinction of IPMN-derived carcinoma and concomitant PDAC for 13 cases where the continuity of adenoma and carcinoma lesions was inconclusive on radiological and pathologic examinations and tissue samples were available, we assessed mutation data on GNAS and KRAS in formalin-fixed paraffin-embedded tissue samples of surgically resected IPMNs and carcinomas (Supplementary Table S2). As described previously (13), we microdissected the intraductal and invasive components on hematoxylin-stained sections and extracted genomic DNA. We conducted PCR and pyrosequencing to assess GNAS and KRAS mutation status (13). We classified GNAS-mutant carcinoma lesions as IPMN-derived. For GNAS-wild-type carcinoma cases, we compared KRAS mutation profiles between carcinoma and IPMN lesions: that is, consistent KRAS mutation patterns may suggest that the carcinoma has been derived from IPMN, and inconsistent patterns may suggest that the carcinoma has evolved through the de novo pathway.

Our primary hypothesis testing was an assessment of the association of ABO blood group with incidence of pancreatic carcinoma overall. Cumulative incidence functions (CIF) of pancreatic carcinoma were estimated on the basis of a competing risk analysis (33, 34). Competing risk events included death without a diagnosis of pancreatic carcinoma as well as occurrence of IPMN-derived or concomitant carcinoma for analyses of the other carcinoma type. The follow-up time was defined as a period from diagnosis of IPMN at our institution to diagnosis of pancreatic carcinoma or the last follow-up, whichever came first. Patients who lost to follow-up were dealt as censored cases at the time of the last follow-up. Using pshreg macro (35), subdistribution hazard ratios (SHR), and 95% CIs were calculated on the basis of the proportional hazards model for competing risks (36). The multivariable proportional hazards model initially included the following variables that were evaluated at baseline: age (continuous), sex (female vs. male), year of diagnosis (continuous), CEA (continuous, log-transformed), CA19–9 (continuous, log-transformed), location of IPMNs (head vs. body-tail vs. multifocal), diameter of the MPD (continuous), size of IPMN (defined as the maximum diameter of the largest cyst, continuous), and number of IPMNs (1 vs. 2–5 vs. ≥ 6). To increase normality of the distribution, we entered log-transformed values for CEA and CA19–9. A backward elimination with a threshold P of 0.05 was conducted to select variables for the final models. For cases with missing data on CEA and CA19–9 (4.9% and 5.7%, respectively), we assigned the median value of each covariate. We confirmed that excluding cases with missing data did not change our results substantially. Using subtype macro (37), we conducted the likelihood ratio test to examine whether the association of ABO blood group with incidence of pancreatic carcinoma differed by carcinoma types, and presented statistical significance as P_{heterogeneity}. We validated the assumption of proportionality of hazards using correlation tests of Schoenfeld residuals and follow-up time (35), and observed evidence on violation of the assumption in the hazard comparing blood groups B and O (P < 0.05). However, the plot of Schoenfeld residuals over time supported the hazard proportionality up to 12 years of the follow-up period, and thus, we used competing-risks proportional hazards models limiting the follow-up period to 12 years. We also performed analyses stratified by tumor characteristics of IPMNs (the MPD diameter, the size of IPMN, or the number of IPMNs). We assessed statistical interaction using the Wald test for the cross-product of ABO blood group and a stratification variable. Given limited numbers of carcinoma cases in several strata, we categorized blood groups into O versus non-O in analyses of carcinoma types and those stratified by IPMN characteristics.

To compare characteristics between blood groups, we used the ANOVA for continuous variables and the χ² test or Fisher exact test, as appropriate, for categorical variables.

All statistical analyses were performed using SAS software (version 9.4, SAS Institute). Two-sided P < 0.05 were considered statistically significant in all analyses.

Table 1 summarizes clinical characteristics of 1,815 patients at diagnosis of IPMNs according to ABO blood groups. In the total study population, ABO blood groups were O in 495 patients (27%), A in 770 patients (42%), B in 356 patients (20%), and AB in 194 patients (11%); this distribution was similar to the data reported in independent cohorts in Japan (38–40). The median diameter of the MPD was 2.0 mm (range, 0.4–20.0 mm), and the median IPMN size was 13 mm (range, 2–74 mm). The rates of main-duct IPMNs (with the MPD ≥ 5 mm) were comparable between the groups.

During 11,518 person-years of follow-up with median duration of 5.2 years (range, 0.5–23.2 years), we documented 97 patients diagnosed with pancreatic carcinoma, including 53 patients with IPMN-derived carcinoma (invasive carcinoma in 37 and high-grade dysplasia in 16) and 44 patients with concomitant PDAC. In the total study population, 1,731 (95%), 1,329 (73%), and 991 (55%) patients were followed for at least 1, 3, and 5 years, respectively. In 69 patients who underwent surgical resection of presumable IPMNs for any indications, the pathologic diagnosis was IPMN-derived carcinoma or concomitant PDAC in 56 patients, IPMN with low-grade dysplasia in 12 patients, and nonneoplastic cyst in 1 patient. Table 2 summarizes clinical characteristics of the patients with pancreatic carcinoma at carcinoma diagnosis, overall and by blood groups. We observed no substantial differences in these clinical parameters between the groups. The diagnosis of pancreatic carcinoma was pathologically confirmed in 85 patients (88%), and all cancerous lesions were classified pathologically as adenocarcinomas (classical ductal adenocarcinoma or IPMN-derived adenocarcinoma).

In our primary hypothesis testing, ABO blood group was associated with the risk of pancreatic carcinoma (Table 3). Compared with patients with blood group O, patients with blood groups A, B, and AB had multivariable SHRs for pancreatic carcinoma of 2.25 (95% CI, 1.25–4.07; P = 0.007), 2.09 (95% CI, 1.08–4.05; P = 0.028), and 1.17 (95% CI, 0.43–3.19; P = 0.76), respectively. The multivariable SHR for pancreatic carcinoma comparing non-O versus O blood groups was 2.02 (95% CI, 1.15–3.55). In a sensitivity analysis limited to 1,560 patients with available data on serologically confirmed ABO blood group, our findings remained largely unchanged: compared with patients with blood group O, patients with blood groups A, B, and AB had multivariable SHRs for pancreatic carcinoma of 2.27 (95% CI, 1.26–4.09), 2.00 (95% CI, 1.03–3.89), and 1.20 (95% CI, 0.45–3.25), respectively. We conducted another sensitivity analysis limited to 85 pathologically confirmed carcinomas as outcomes, which yielded similar results: compared with patients with blood group O, patients with blood groups A, B, and AB had multivariable SHRs for pancreatic carcinoma of 2.15 (95% CI, 1.14–4.07), 2.03 (95% CI, 1.00–4.11), and 1.24 (95% CI, 0.45–3.40), respectively. Given slow progression of a majority of IPMNs, we conducted a sensitivity analysis limited to 1,731 patients followed for more than one year, which yielded similar results: compared with patients with blood group O, patients with blood groups A, B, and AB had multivariable SHRs for pancreatic carcinoma of 2.30 (95% CI, 1.27–4.17), 2.08 (95% CI, 1.07–4.03), and 1.18 (95% CI, 0.43–3.20), respectively. CIF curves of pancreatic carcinoma, overall and by carcinoma types, are illustrated in Fig. 2; Supplementary Fig. S1, respectively. Pancreatic cancer-free survival rates at 2, 5, and 10 years of IPMN diagnosis were as follows: 98.6%, 98.3%, and 95.1%, respectively, for blood group O; 97.5%, 94.2%, and 91.7%, respectively, for blood group A; 97.8%, 95.4%, and 90.2%, respectively, for blood group B; and 97.7%, 96.3%, and 93.9%, respectively, for blood group AB. When we examined specific carcinoma types, a modestly increased risk of pancreatic carcinoma associated with non-O blood group was consistently observed both in IPMN-derived carcinoma and concomitant PDAC (P_{heterogeneity} = 0.65, Supplementary Table S3), but statistical power was limited. Considering the possibility of preclinical malignancy, we conducted a sensitivity analysis excluding cases with IPMNs harboring a mural nodule at diagnosis (as one of high-risk stigmata; ref. 19), which yielded consistent results. Main-duct IPMNs were associated with higher risk of pancreatic carcinoma compared with branch-duct IPMNs (SHR, 4.34; 95% CI, 2.52–7.47).

When we examined the association of ABO blood group with the risk of pancreatic carcinoma stratified by cyst characteristics (the MPD diameter, the size of IPMN, or the number of IPMNs), we did not observe any effect modification by these characteristics (P_interaction > 0.39, Table 4). Compared with patients with blood group O and branch-duct IPMN (with the MPD < 5 mm), patients with non-O blood group and main-duct IPMN (with the MPD ≥ 5 mm) had multivariable SHR of 9.93 (95% CI, 4.28–23.1).

In this large longitudinal cohort study, IPMNs occurring among individuals with blood group A or B appeared to represent higher potential of progressing to pancreatic carcinoma compared with those occurring among individuals with blood group O. The elevated risk was consistently observed in IPMN-derived and concomitant ductal carcinomas, suggesting similar contributions of the blood group antigens to well-documented distinct pathways of pancreatic carcinogenesis. The high risk of pancreatic cancer associated with non-O blood group appeared to be consistent across strata of major morphologic characteristics of IPMNs. Notably, patients with main-duct IPMN with non-O blood group had approximately 10-fold higher risk of pancreatic carcinoma compared with patients with branch-duct IPMN with blood group O. Our data suggest that information on ABO blood group may help to mark a subpopulation of patients with IPMNs at particularly high risk of pancreatic carcinogenesis, potentially providing insights into management strategies of patients with IPMN.

ABO blood group is inherently determined by alleles of the ABO gene, which encodes a glycosyltransferase that catalyzes the transfer of carbohydrates to a protein backbone (the H antigen) and forms red blood cell antigens. The antigens express not only on red blood cells but also gastrointestinal epithelial cells (41), and cancerous and noncancerous pancreatic tissues in the human (42, 43). Mechanistic evidence supports that the glycosyltransferase levels may be involved in alterations in intercellular adhesion, cellular membrane signaling, and/or inflammatory and immune responses (44, 45). In an expression quantitative trait locus (eQTL) analysis (46), a SNP at the ABO gene locus may influence ABO expression status in the pancreas. In a prospective study based on large cohorts in the United States, non-O blood group (A, B, or AB) was associated with 1.3- to 1.7-fold higher incidence of pancreatic cancer compared with blood group O (3). These findings were subsequently validated by a meta-analysis involving 24 cohort studies with summary ORs of 1.40 (95% CI, 1.28–1.53), 1.19 (95% CI, 1.05–1.35), and 1.29 (95% CI, 1.10–1.51) for blood groups A, B, and AB (vs. O), respectively (2). In a genome-wide association study, a lower risk of pancreatic cancer was noted for individuals harboring the allele T for the SNP rs505922, which was in complete linkage disequilibrium with the O allele of the ABO locus (5). ABO blood group is a readily measured marker that is not susceptible to other endogenous and exogenous factors, potentially facilitating simple risk prediction for the aggressive malignancy.

Our findings suggest that ABO blood group may be associated with the risk of pancreatic carcinogenesis in patients with IPMNs, supporting the potential of ABO blood group as a biomarker for risk stratification of this patient population. The risk associated with non-O blood group appeared to be further elevated when the MPD was dilatate. In an analysis of genotype-derived allelic variations in ABO blood group (i.e., OO, AO, AA, BO, BB, and AB), the researchers observed an increased risk of pancreatic cancer because of the addition of each non-O allele in general U.S. populations (47). Furthermore, a specific blood group A allele associated with elevated glycosyltransferase activity appeared to further increase the risk (48). Therefore, an investigation incorporating genotyping data is warranted to support our findings of higher risk of pancreatic cancer associated with non-O blood group among patients with IPMNs. In a cross-sectional analysis of large surgical series in the United States (49), non-O blood groups were more frequently noted in patients with IPMN than in general U.S. populations, but were associated with a lower likelihood of high-grade dysplasia, suggesting heterogeneous effects of the blood group antigens on development and progression of IPMNs. In contrast to this study, our study examined ABO blood groups in relation to progression of IPMNs to pancreatic carcinoma during long-term surveillance. Within a longitudinal cohort study (50), the researchers analyzed data on 89 patients to examine genotyped ABO blood group in relation to progression of IPMNs and found that AA genotype might be associated with the progression compared with genotype OO. However, this analysis was limited by the small sample size and the lack of consideration of cancer development as an outcome variable.

Major pathogenic processes implicated in carcinogenesis associated with IPMNs include carcinoma derived from IPMN and de novo pancreatic ductal adenocarcinoma (51, 52). Main genetic drivers for PDAC have been mutations in KRAS, CDKN2A (p16), TP53 (p53), and SMAD4 (53), which are also observed in IPMNs with lower frequency (54, 55). IPMN-derived carcinomas are characterized by other molecular alterations including mutations in GNAS and RNF43 (54–56). In particular, GNAS mutation has been noted specifically in IPMNs and IPMN-derived carcinomas, but rarely in de novo PDACs or other cystic neoplasms of the pancreas (54, 55). On the basis of the recognition of the heterogeneous pathways of pancreatic carcinogenesis in patients with IPMNs, our data have suggested that the relationship of ABO blood group with pancreatic cancer incidence may not differ significantly by the carcinogenic processes. A functional study is warranted to elucidate the mechanism through which non-O blood group alleles drive the development of both IPMN-derived and de novo ductal adenocarcinomas among patients with IPMNs.

This study has notable strengths including the use of our large longitudinal data on patients with IPMN with a follow-up period of up to 25 years. The large sample size with close to 100 pancreatic carcinoma cases allowed us to elucidate the clinical course of patients diagnosed with IPMNs according to blood groups. Along with meticulous pathologic examinations, molecular analyses of paired carcinoma and IPMN tissue samples refined our differential diagnoses of IPMN-derived carcinoma and concomitant PDAC for inconclusive cases. Our study utilized the molecular pathologic epidemiology (MPE) approach to examine differential associations of ABO blood group with incidence of pancreatic carcinoma by carcinoma subtypes (57, 58). The integrative discipline of MPE examines differential contributions of an exposure of interest to incidences of tumors subclassified by molecular pathologic features and potentially provides evidence for carcinogenesis driven by specific pathways in human populations.

This study has limitations. First, we analyzed data on patients diagnosed with IPMNs at a single tertiary care center, potentially resulting in a referral bias. Second, data on ABO blood group were not available for all patients diagnosed with IPMNs; however, clinical characteristics were comparable between patients with and without the data. The patients diagnosed with pancreatic carcinoma were more likely to undergo testing for ABO blood group, potentially contributing to a higher crude rate of pancreatic carcinoma in this study population. However, ABO blood group less likely affected the availability of the data. Third, the relatively small number of patients with blood group AB precluded a robust statistical assessment in this stratum. Fourth, there is a possibility of unmeasured confounding factors. We did not adjust for well-known risk factors for pancreatic cancer in the general population (e.g., smoking, obesity, diabetes, heavy alcohol drinking; refs. 59, 60). Nonetheless, the adjustment for major risk factors for pancreatic carcinogenesis reported in patients with IPMN including the MPD dilatation did not alter our findings substantially. In addition, ABO blood group is inherently determined, and therefore, epidemiologic factors were less likely to be confounding factors in the association of ABO blood group with pancreatic cancer development. Finally, a vast majority of the study population was Japanese. Therefore, our data should be validated in independent populations.

In conclusion, patients with IPMN with non-O blood group appeared to be at higher risk of pancreatic carcinoma compared with patients with blood group O. Given that ABO blood group is a readily available biomarker, our findings, if validated, would facilitate risk stratification of patients diagnosed with IPMNs and thereby refine surveillance strategies. Further research is warranted to explore the mechanism by which genetic variants at the ABO gene promote malignant transformation of pancreatic cells in this population.

K. Hasegawa reports grants from Taiho Pharmacia, Chugai Pharmacia, Takeda Pharmacia, and Bayer Pharmacia outside the submitted work. No disclosures were reported by the other authors.

The funders had no role in study design, data collection, and analysis; decision to publish; or preparation of the manuscript.

T. Hamada: Conceptualization, resources, data curation, software, formal analysis, funding acquisition, investigation, visualization, methodology, writing–original draft. H. Oyama: Conceptualization, resources, data curation, software, formal analysis, investigation, visualization, methodology, writing–original draft. Y. Nakai: Resources, data curation, supervision, writing–review and editing. M. Tada: Resources, data curation, writing–review and editing. H. Koh: Validation, methodology, writing–review and editing. K. Tateishi: Resources, data curation, supervision, writing–review and editing. J. Arita: Resources, writing–review and editing. R. Hakuta: Resources, data curation. H. Ijichi: Resources, data curation, writing–review and editing. K. Ishigaki: Resources, data curation. Y. Kawaguchi: Resources, writing–review and editing. H. Kogure: Resources, data curation. S. Mizuno: Resources, data curation. T. Morikawa: Resources, writing–review and editing. K. Saito: Resources, data curation. T. Saito: Resources, data curation. T. Sato: Resources, data curation. K. Takagi: Resources, data curation, investigation. N. Takahara: Resources, data curation. R. Takahashi: Resources, data curation, writing–review and editing. A. Tanaka: Resources, data curation, writing–review and editing. M. Tanaka: Resources, data curation, investigation, writing–review and editing. T. Ushiku: Resources, writing–review and editing. K. Hasagawa: Resources, writing–review and editing. K. Koike: Supervision, project administration, writing–review and editing.

The authors thank the following contributors for their valuable support in data collection: Akiko Kunita and Yasuyuki Morishita, The University of Tokyo (Tokyo, Japan); Takeyuki Watadani and Osamu Abe, The University of Tokyo (Tokyo, Japan); Hiroyuki Isayama, Juntendo University (Tokyo, Japan); and Kazumichi Kawakubo, Hokkaido University Hospital (Hokkaido, Japan). This work was supported by grants from Japan Society for the Promotion of Science (JSPS) KAKENHI (JP19K08362, to T. Hamada), The Yasuda Medical Foundation (to T. Hamada), and Pancreas Research Foundation of Japan (to T. Hamada).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.