Background:

On the basis of a previous report of increased chronic myeloid leukemia (CML) risk following peptic ulcer, we hypothesized that chronic Helicobacter pylori infection could serve as a risk factor for CML.

Methods:

In a population-based, retrospective case–control study, we used Swedish registry data on 980 patients with CML and 4,960 age- and sex-matched controls to investigate associations between markers of previous infection with Helicobacter pylori and CML incidence.

Results:

Previous diagnoses of dyspepsia, gastritis or peptic ulcers, as well as previous proton pump inhibitor (PPI) medication, were all associated with a significantly increased risk of CML (RRs, 1.5–2.0; P = 0.0005–0.05). Meanwhile, neither inflammatory bowel disease nor intake of NSAIDs were associated with CML, indicating that it is not gastrointestinal ulcer or inflammation per se that influences risk.

Conclusions:

The consistent associations suggest a shared background between gastric conditions and CML, and strengthen the case that Helicobacter pylori could constitute this common risk factor.

Impact:

As the etiology of CML is practically unknown, and Helicobacter pylori could potentially be a therapeutic target, even this indirect evidence encourages further studies on the potential involvement of Helicobacter pylori in CML etiology.

Sporadic reports on peptic ulcers in patients with chronic myeloid leukemia (CML) have appeared in the scientific literature at least since the mid-seventies (1–5). The proposed explanations for this observed association have varied; some have speculated in histamine release by basophils while others have suggested platelet dysfunction, but it has generally been assumed that any increased risk of ulcers would be a consequence of the hematologic disease. However, in 2012 a study was published investigating possible risk factors for myeloid malignancies, including 186 cases of CML (6). In this study, an increased incidence of ulcers before onset of CML was noted. As the investigators ruled out aspirin use as a confounder, they instead proposed chronic Helicobacter pylori infection as a common risk factor for both the observed ulcers and the CML.

Helicobacter pylori is classified as a carcinogen since 1994, and is associated with both gastric cancer (7) and mucosa-associated lymphoid tissue (MALT) lymphoma (8). Interestingly, a growing body of evidence indicates that the cancer stem cells in both of these conditions may be of hematopoietic origin. Thus, although gastric cancer appears in the gastric mucosa epithelium, several observations suggest that the cancer cells may originate from circulating bone marrow stem cells, recruited to the mucosa to replace normal gastric stem cells that have underwent apoptosis due to Helicobacter pylori infection (7, 9).

To study the potential association between Helicobacter pylori infection and the risk of developing CML, we performed a population-based case–control study using several different Swedish patient registries. We compared the nearly 1,000 Swedish CML cases diagnosed between 2002 and 2012 with nearly 5,000 age- and sex-matched controls. As the study was based on registry data, we had no direct evidence of Helicobacter infections in the patients and controls. Instead, as proxy variables we used documented history of related medical conditions, such as peptic ulcer and prescribed medication with proton pump inhibitors (PPI).

Data sources

Cases were all patients registered in the Swedish CML Register with a diagnosis of CML (ICD10 = C92.1) between January 2002 and December 2012. The CML register was initiated in 2002 and includes approximately 97% of all Swedish patients with CML as evaluated by mandatory pathology reports recorded in the Swedish Cancer Register (10). In essentially all patients (>98%), the CML diagnosis was confirmed by the cytogenetic presence of a Philadelphia chromosome or the molecular finding of a BCR-ABL1 transcript. The CML register also comprises clinical and laboratory data obtained at diagnosis and from regular subsequent follow-up assessments performed at least yearly or biyearly. From data reported at diagnosis, risk groups according to Sokal and colleagues (11) or Hasford and colleagues (12) can be calculated.

Matched to each CML case, five population control subjects were randomly chosen from the Swedish Total Population Register. The control subjects were of the same sex, born in the same year, and alive and resident in the same region as the respective patient with CML, as assessed in the beginning of the year during which the patient was diagnosed. Control subjects who already had a CML diagnosis or died between January 1 and the date of diagnosis were excluded from analysis. Both patients and control subjects were followed for death or permanent emigration, as documented in the Swedish Total Population Register until December 31, 2013.

Data on PPI treatment and treatment with antibiotics commonly used for Helicobacter pylori eradication preceding CML diagnosis were retrieved from the Swedish Prescribed Drug Register, by means of record linkage. Potential confounders such as treatment with acetylsalicylic acid (ASA) or NSAID were also retrieved. The Prescribed Drug Register was established in 2005 and includes all dispensed prescriptions in the Swedish population (13). To avoid reverse causality, only medications prescribed at least 12 months before date of diagnosis were included in the analysis.

Records of diagnoses potentially related to Helicobacter pylori, such as peptic ulcer and gastritis, and presenting before a documented CML diagnosis, were retrieved from the National Patient Register, which holds records of ICD-coded hospital discharge diagnoses. As with medications, diagnoses set less than 12 months before the CML diagnosis were excluded. The National Patient Register covers all Swedish in-patient care from 1987 and onwards. Since 2001, also diagnoses from specialized out-patient care are recorded (14). Comorbidity was estimated using the Charlson comorbidity index (15), including all relevant diagnoses from the register during 10 years preceding CML diagnosis.

To assess potential confounding, socioeconomic characteristics were obtained from the integrated database for labor market research (LISA; ref. 16). It contains data from 1990 onwards on education, income, sick leave, and several other socioeconomic indicators for all Swedish citizens 16 years of age and older. The LISA database is a compilation of data from several registers at Statistics Sweden and other authorities.

Statistical analysis

All analyses were performed with SAS 9.4 Statistical Software (SAS Inc.). As all controls were individually matched on follow-up time, the design can be viewed as a nested case–control study with the population of Sweden as the underlying cohort. In case–control comparisons (presented in Tables 1 and 2), the data were accordingly analyzed with risk set stratified Cox regression (PHREG in SAS), and RRs were estimated by hazard ratios (HR) with 95% confidence intervals (95% CI). To be included in each analysis, exposure data needed to be complete for the case and at least one corresponding control. As potential confounders, only baseline characteristics that had shown a significant association to CML incidence in univariate models (Table 1) were included in the multivariable regression models. These included education level, living alone and comorbidity index. Disease characteristics comparisons between groups (patients with or without gastric conditions, Table 3) were analyzed using a generalized linear model (17). One- and 5-year survival were estimated by the actuarial approach using PROC LIFETEST in SAS (18), and relative survival between groups was analyzed by Cox regression (ref. 19; Table 4). All P values were two-sided and a value below 0.05 was considered statistically significant.

Table 1.

Characteristics of 980 cases of CML diagnosed in Sweden between 2002 and 2012 and 4,860 controls matched on year of birth, sex, and region of residence.

CasesControls
Characteristicn (%)n (%)RR (95% CI)P
Age at diagnosis   MC — 
 <45 232 (24) 1,161 (24)   
 45–59 250 (26) 1,242 (26)   
 60–74 310 (32) 1,538 (32)   
 ≥75 188 (19) 919 (19)   
Sex   MC — 
 Male 539 (55) 2,671 (55)   
 Female 441 (45) 2,189 (45)   
Personal yearly income     
 <97,200 194 (20) 972 (20) Ref  
 97,200–126,100 194 (20) 972 (20) 1.00 (0.80–1.25) 1.0 
 126,200–160,700 206 (21) 972 (20) 1.06 (0.85–1.33) 0.6 
 160,800–221,300 212 (22) 971 (20) 1.09 (0.87–1.37) 0.4 
 ≥221,400 174 (18) 973 (20) 0.88 (0.69–1.13) 0.3 
Household yearly income     
 <141,000 190 (19) 972 (20) Ref — 
 141,000–217,600 194 (20) 971 (20) 1.03 (0.83–1.29) 0.8 
 217,700–313,100 185 (19) 973 (20) 0.99 (0.79–1.25) 1.0 
 313,200–451,400 205 (21) 973 (20) 1.11 (0.88–1.41) 0.4 
 ≥451,500 206 (21) 971 (20) 1.13 (0.89–1.44) 0.3 
Education (years in school)a     
 ≤9 327 (34) 1,535 (32) Ref — 
 10–12 411 (43) 1,948 (41) 0.98 (0.83–1.16) 0.8 
 >12 227 (24) 1,299 (27) 0.80 (0.66–0.98) 0.03 
Living alone     
 Yes 332 (34) 1,810 (37) Ref — 
 No 648 (66) 3,050 (63) 0.85 (0.73–0.98) 0.03 
Charlson comorbidity index     
 0 742 (76) 3,962 (82) Ref — 
 1 112 (11) 377 (8) 1.69 (1.33–2.15) <0.0001 
 2 80 (8) 309 (6) 1.46 (1.11–1.90) 0.006 
 ≥3 46 (5) 212 (4) 1.25 (0.89–1.76) 0.2 
CasesControls
Characteristicn (%)n (%)RR (95% CI)P
Age at diagnosis   MC — 
 <45 232 (24) 1,161 (24)   
 45–59 250 (26) 1,242 (26)   
 60–74 310 (32) 1,538 (32)   
 ≥75 188 (19) 919 (19)   
Sex   MC — 
 Male 539 (55) 2,671 (55)   
 Female 441 (45) 2,189 (45)   
Personal yearly income     
 <97,200 194 (20) 972 (20) Ref  
 97,200–126,100 194 (20) 972 (20) 1.00 (0.80–1.25) 1.0 
 126,200–160,700 206 (21) 972 (20) 1.06 (0.85–1.33) 0.6 
 160,800–221,300 212 (22) 971 (20) 1.09 (0.87–1.37) 0.4 
 ≥221,400 174 (18) 973 (20) 0.88 (0.69–1.13) 0.3 
Household yearly income     
 <141,000 190 (19) 972 (20) Ref — 
 141,000–217,600 194 (20) 971 (20) 1.03 (0.83–1.29) 0.8 
 217,700–313,100 185 (19) 973 (20) 0.99 (0.79–1.25) 1.0 
 313,200–451,400 205 (21) 973 (20) 1.11 (0.88–1.41) 0.4 
 ≥451,500 206 (21) 971 (20) 1.13 (0.89–1.44) 0.3 
Education (years in school)a     
 ≤9 327 (34) 1,535 (32) Ref — 
 10–12 411 (43) 1,948 (41) 0.98 (0.83–1.16) 0.8 
 >12 227 (24) 1,299 (27) 0.80 (0.66–0.98) 0.03 
Living alone     
 Yes 332 (34) 1,810 (37) Ref — 
 No 648 (66) 3,050 (63) 0.85 (0.73–0.98) 0.03 
Charlson comorbidity index     
 0 742 (76) 3,962 (82) Ref — 
 1 112 (11) 377 (8) 1.69 (1.33–2.15) <0.0001 
 2 80 (8) 309 (6) 1.46 (1.11–1.90) 0.006 
 ≥3 46 (5) 212 (4) 1.25 (0.89–1.76) 0.2 

Abbreviation: MC, matching criterion.

aMissing data on 15 cases and 78 controls.

Table 2.

RR of CML following a number of previous diagnoses and historic use of prescription drugs.

CasesControlsb
Characteristican (%)n (%)Crude RR (95% CI)PAdjusted RRc (95% CI)P
Diagnoses (ICD10) 
 Dyspepsia (K30) 20 (2.0) 51 (1.1) 2.0 (1.2–3.3) 0.01 1.8 (1.1–3.1) 0.03 
 Gastritis (K21) 34 (3.5) 96 (2.0) 1.8 (1.2–2.7) 0.004 1.6 (1.1–2.4) 0.02 
 Peptic ulcer (K25-K27) 31 (3.2) 87 (1.8) 1.8 (1.2–2.7) 0.006 1.7 (1.1–2.6) 0.02 
 Gastric ulcer (K25) 13 (1.3) 41 (0.8) 1.6 (0.8–2.9) 0.2 1.5 (0.8–2.7) 0.2 
 Duodenal ulcer (K26) 18 (1.8) 45 (0.9) 2.0 (1.2–3.5) 0.01 1.9 (1.1–3.3) 0.03 
 Inflammatory bowel disease (K50-52) 16 (1.6) 75 (1.5) 1.1 (0.6–1.8) 0.8 1.1 (0.6–1.8) 0.8 
Medications (ATC) 
 PIPs (A02BC) 133 (13.6) 462 (9.5) 1.6 (1.3–2.0) <0.0001 1.5 (1.2–1.9) 0.0005 
 NSAIDs (M01A) 202 (20.6) 1,007 (20.7) 1.0 (0.8–1.2) 0.9 1.0 (0.8–1.2) 0.9 
 ASA (B01AC06, N02BA01, N02BA51) 120 (12.2) 531 (10.9) 1.2 (0.9–1.5) 0.2 1.1 (0.9–1.4) 0.5 
 Glucocorticoids (H02AB) 66 (6.7) 282 (5.8) 1.2 (0.9–1.6) 0.3 1.1 (0.8–1.5) 0.4 
CasesControlsb
Characteristican (%)n (%)Crude RR (95% CI)PAdjusted RRc (95% CI)P
Diagnoses (ICD10) 
 Dyspepsia (K30) 20 (2.0) 51 (1.1) 2.0 (1.2–3.3) 0.01 1.8 (1.1–3.1) 0.03 
 Gastritis (K21) 34 (3.5) 96 (2.0) 1.8 (1.2–2.7) 0.004 1.6 (1.1–2.4) 0.02 
 Peptic ulcer (K25-K27) 31 (3.2) 87 (1.8) 1.8 (1.2–2.7) 0.006 1.7 (1.1–2.6) 0.02 
 Gastric ulcer (K25) 13 (1.3) 41 (0.8) 1.6 (0.8–2.9) 0.2 1.5 (0.8–2.7) 0.2 
 Duodenal ulcer (K26) 18 (1.8) 45 (0.9) 2.0 (1.2–3.5) 0.01 1.9 (1.1–3.3) 0.03 
 Inflammatory bowel disease (K50-52) 16 (1.6) 75 (1.5) 1.1 (0.6–1.8) 0.8 1.1 (0.6–1.8) 0.8 
Medications (ATC) 
 PIPs (A02BC) 133 (13.6) 462 (9.5) 1.6 (1.3–2.0) <0.0001 1.5 (1.2–1.9) 0.0005 
 NSAIDs (M01A) 202 (20.6) 1,007 (20.7) 1.0 (0.8–1.2) 0.9 1.0 (0.8–1.2) 0.9 
 ASA (B01AC06, N02BA01, N02BA51) 120 (12.2) 531 (10.9) 1.2 (0.9–1.5) 0.2 1.1 (0.9–1.4) 0.5 
 Glucocorticoids (H02AB) 66 (6.7) 282 (5.8) 1.2 (0.9–1.6) 0.3 1.1 (0.8–1.5) 0.4 

Abbreviations: ATC, Anatomic Therapeutic Chemical classification system; ICD, International Classification of Diseases.

aVariables are not mutually exclusive; a subject could, for example, have both a previous peptic ulcer and a previous PPI prescription.

bMatched on sex, age, and region of residence.

cAdjusted for education level, living alone, and comorbidity index (missing data on 15 cases and 78 controls).

Table 3.

Clinical characteristics at diagnosis of patients with CML with a previous documented history of peptic ulcer, gastritis, or dyspepsia or with previous prescribed PPI use (Group A) compared with CML patients without such history (Group B).

Group AGroup B
n (%)n (%)Ptrend
Age category 
 <45 years 21 (12) 211 (26)  
 45–59 years 35 (21) 215 (27)  
 60–74 years 67 (40) 243 (30)  
 ≥75 years 46 (25) 142 (18)  
   <0.0001 
Sex 
 Male 99 (59) 440 (54)  
 Female 70 (41) 371 (46)  
   0.3 
WHO performance status 
 0 104 (64) 455 (59)  
 1 43 (26) 242 (31)  
 2 11 (7) 47 (6)  
 3 4 (2) 22 (3)  
 4 1 (1) 6 (1)  
   0.8 
Phase at presentation 
 Chronic 159 (94) 744 (93)  
 Accelerated 6 (4) 36 (5)  
 Blast crisis 4 (2) 20 (3)  
   0.9 
Sokal risk group 
 Low risk (<0.8) 33 (20) 175 (23)  
 Intermediate (0.8–1.2) 85 (52) 311 (40)  
 High risk (>1.2) 46 (28) 282 (37)  
   0.02 
Hasford risk group 
 Low risk (≤780) 46 (28) 241 (32)  
 Intermediate (781–1,480) 96 (59) 360 (47)  
 High risk (>1,480) 21 (13) 161 (21)  
   0.01 
Basophil percentage at diagnosisa 
 <3% 62 (37) 301 (38)  
 ≥3% 104 (63) 488 (62)  
    0.8 
Eosinophil percentage at diagnosisb 
 <3% 112 (68) 461 (59)  
 ≥3% 54 (33) 327 (42)  
   0.03 
Blast percentage at diagnosisc 
 <1% 97 (59) 347 (44)  
 ≥1% 67 (41) 437 (56)  
   0.0005 
Group AGroup B
n (%)n (%)Ptrend
Age category 
 <45 years 21 (12) 211 (26)  
 45–59 years 35 (21) 215 (27)  
 60–74 years 67 (40) 243 (30)  
 ≥75 years 46 (25) 142 (18)  
   <0.0001 
Sex 
 Male 99 (59) 440 (54)  
 Female 70 (41) 371 (46)  
   0.3 
WHO performance status 
 0 104 (64) 455 (59)  
 1 43 (26) 242 (31)  
 2 11 (7) 47 (6)  
 3 4 (2) 22 (3)  
 4 1 (1) 6 (1)  
   0.8 
Phase at presentation 
 Chronic 159 (94) 744 (93)  
 Accelerated 6 (4) 36 (5)  
 Blast crisis 4 (2) 20 (3)  
   0.9 
Sokal risk group 
 Low risk (<0.8) 33 (20) 175 (23)  
 Intermediate (0.8–1.2) 85 (52) 311 (40)  
 High risk (>1.2) 46 (28) 282 (37)  
   0.02 
Hasford risk group 
 Low risk (≤780) 46 (28) 241 (32)  
 Intermediate (781–1,480) 96 (59) 360 (47)  
 High risk (>1,480) 21 (13) 161 (21)  
   0.01 
Basophil percentage at diagnosisa 
 <3% 62 (37) 301 (38)  
 ≥3% 104 (63) 488 (62)  
    0.8 
Eosinophil percentage at diagnosisb 
 <3% 112 (68) 461 (59)  
 ≥3% 54 (33) 327 (42)  
   0.03 
Blast percentage at diagnosisc 
 <1% 97 (59) 347 (44)  
 ≥1% 67 (41) 437 (56)  
   0.0005 

Abbreviation: WHO, World Health Organization.

aMissing data for 25 patients (3%).

bMissing data for 25 patients (3%).

cMissing data for 32 patients (3%).

Table 4.

Survival of patients with CML with (Group A) or without (Group B) a previous documented history of peptic ulcer, gastritis, or dyspepsia, or use of PPIs.

Unadjusted dataAge adjusted data
n1-Year survival5-Year survivalHR (95% CI)HR (95% CI)
Group A 169 89 78 1.2 (0.9–1.8) 0.8 (0.6–1.2) 
 Peptic ulcer 31 71 59 2.4 (1.4–4.2) 1.9 (1.1–3.3) 
 Gastritis 34 91 64 1.7 (0.9–3.2) 1.3 (0.7–2.4) 
 Dyspepsia 20 90 90 0.8 (0.3–2.5) 0.5 (0.2–1.5) 
 PPI 133 92 81 1.0 (0.7–1.6) 0.7 (0.4–1.1) 
Group B 811 94 80 1 (ref; —) 1 (ref; —) 
Unadjusted dataAge adjusted data
n1-Year survival5-Year survivalHR (95% CI)HR (95% CI)
Group A 169 89 78 1.2 (0.9–1.8) 0.8 (0.6–1.2) 
 Peptic ulcer 31 71 59 2.4 (1.4–4.2) 1.9 (1.1–3.3) 
 Gastritis 34 91 64 1.7 (0.9–3.2) 1.3 (0.7–2.4) 
 Dyspepsia 20 90 90 0.8 (0.3–2.5) 0.5 (0.2–1.5) 
 PPI 133 92 81 1.0 (0.7–1.6) 0.7 (0.4–1.1) 
Group B 811 94 80 1 (ref; —) 1 (ref; —) 

Ethics approval

The study was approved by the Uppsala regional research ethics committee (2013/069). Record linkages were performed at Statistics Sweden and the National Board of Health and Welfare. All data were de-identified before delivery to the researchers. For this type of study individual consent is not required.

Between 2002 and 2012, 984 individuals diagnosed with CML were included in the Swedish CML register. Four subjects were excluded from the analysis, because they were not identifiable in Swedish population statistics. All four were non-EU citizens and presumably only on a temporary stay in Sweden when diagnosed with CML. Furthermore, among the 4,900 matched control subjects, 39 were excluded as they died before matching date, and 1 was excluded because of a record of a prior diagnosis of CML. This way, the final study encompassed 980 patients with CML and 4,860 matched controls.

The median age of the patients at CML diagnosis was 60 (range, 17–99) years, 539 (55%) being men and 441 (45%) women. Compared with the controls, the CML cases had a moderately higher co-morbidity index, a shorter education, and were less likely to live alone (Table 1). Income the year before diagnosis did not differ between cases and controls.

Compared with controls, CML cases had significantly more often a previous diagnosis of dyspepsia, gastritis, or an ulcer of peptic or gastric origin, with adjusted RRs between 1.5 and 1.9 (Table 2). Adjustments for socioeconomic factors and comorbidity did not substantially change these estimates. In contrast, the incidence of inflammatory bowel disease did not vary between cases and controls. Cases were more likely to have had PPI prescribed (adjusted RR 1.5), whereas the previously prescribed use of NSAIDs, ASA, or glucocorticoids did not differ between the cohorts. In total, 17% of cases compared with 12% of controls had an earlier diagnosis of dyspepsia, gastritis, or peptic ulcer documented, or had been prescribed PPIs (P < 0.0001, Table 2). No significant associations were detected between previous use of amoxicillin, tetracycline, metronidazole, or clarithromycin and CML risk, although with low statistical power (Supplementary Table S1).

Among patients with CML, those with a history of prior gastric conditions or PPI medication differed significantly from those without, in that they were older and presented with lower eosinophil and blast cell counts as well as lower Sokal/Hasford risk scores at diagnosis (despite the higher age; Table 3). Neither the performance status, nor the proportion diagnosed in accelerated phase/blast crisis differed between the two groups.

Risk of death during follow-up did not differ between patients with and without previous gastric conditions (Table 4). Patients with previous peptic ulcers did, however, have a doubled risk of death during follow-up, even after adjustment for their older age. A possible explanation could be that individuals with a history of peptic ulcers have poorer health and higher risk of an early death. Indeed, control subjects with a history of peptic ulcers also had a nearly doubled age-adjusted risk of death during follow-up compared with other controls (HR, 1.8; 95% CI, 1.2–2.6), and the relative survival (compared with controls) was similar for CML cases with or without previous peptic ulcer (HR, 1.7 and HR, 2.0, respectively).

In this, to date largest published study, we report that patients with a history of peptic ulcer are at an increased risk of acquiring CML. We further observed a similar pattern for patients with a previous, documented dyspepsia or gastritis, as well as for patients with prior use of PPIs. All of these conditions are markers of populations with increased prevalence of Helicobacter infections (20, 21). Strengthening the case that these correlations are due to underlying Helicobacter infections, no increased risk was noted among users of glucocorticoids, ASA or NSAIDs, which is an alternative etiology for peptic ulcers. In line with previous findings (22, 23), we did not observe an increased risk of CML among patients with inflammatory bowel disease, indicating that it is not gastrointestinal inflammation per se which promotes the CML risk.

Reverse causality, through an increased basophil count predating CML diagnosis and increasing risk of ulcers, is also unlikely as patients with previous gastric conditions or medication were not found to have higher basophil counts at diagnosis than other patients with CML. Moreover, to consider reverse causality as an explanation to our findings it must be assumed that (i) gastrointestinal symptoms of CML would appear more than a year before diagnosis, (ii) these symptoms would be severe enough to warrant medical attention while laboratory tests still do not raise the doctor's suspicion of an underlying CML, and (iii) that this is not a sporadic event but happening to a considerable proportion of patients with CML.

Prescription of amoxicillin, tetracycline, metronidazole, clarithromycin, or combination packs were not associated with CML risk (Supplementary Table S1), but statistical power was low. Only three cases and six controls had been prescribed clarithromycin and combination packs had only been prescribed to 6 patients and 19 controls, suggesting a substantial underreporting. A possible explanation could be that Helicobacter eradication has often been given during in-patient care or been handed out directly by endoscopy units and less often prescribed.

In a previous, smaller study Johnson and colleagues (6) noted an increased self-reported prevalence of peptic ulcers among 186 patients with CML, that is, in-line with our data. Additional reports on gastrointestinal conditions as risk factors for CML are, to our knowledge, lacking. Gunnarsson and colleagues (23, 24) found increases in incidence of gastrointestinal malignancies both before and after diagnosis of CML, but did not specify localization of the tumors.

The comparison of disease characteristics between patients with CML with or without history of gastric conditions (Table 3) should be viewed as exploratory. However, from our data we have no reason to suggest that patients with previous gastric conditions generally suffer from a more aggressive hematologic disease than other patients with CML. Rather, such patients were less likely to be considered high risk at diagnosis (according to Sokal or Hasford), or to have more than 1% circulating blast cells. Hence, our data could indicate that patients with previous gastric conditions may have a somewhat better prognosis, but in our studied cohorts this did not translate into a better age-adjusted survival, keeping in mind that such overall survival comparisons may be influenced by a number of confounding factors we could not control.

This study only provides a statistical link between the history of Helicobacter-related diseases and the risk of CML, leaving the biological mechanisms behind this link open to speculation. Although CML is believed to be caused by a single translocation in a hematopoietic stem cell, epidemiological evidence accrued by Greaves (25), Kinlen (26), and others suggests that infections might be instrumental in the etiology of hematological malignancies, in spite of what we otherwise know about their pathogenesis. It has been argued that the chromosomal change might actually predate the clinical disease, and that the role of infections can be to turn the pre-leukemic clone into overt disease (27). In parallel, both Epstein–Barr virus infection and c-myc translocation take part in the development of endemic Burkitt leukemia (28), whereas cytomegalovirus infections have recently been implicated in childhood acute lymphoblastic leukemia etiology (29, 30). In support of this hypothesis, Helicobacter pylori is also considered causally related to MALT lymphoma of the gastric mucosa, a disease often associated with t(11;18) and less frequently t(1;14) translocations (31). In CML, one possible mode of action could be through dysregulation of the tumor-suppressing miRNA 203 (miR-203). miRNAs are small noncoding RNAs with regulatory functions of gene expression. miR-203 is suppressed by Helicobacter pylori (32), and it has been demonstrated that such suppression enhances expression of ABL1 and BCR-ABL1 (33). miR-203 silencing has also been implicated as a mediator in the relationship between Helicobacter and MALT lymphoma (34). Interestingly, miR-203 function is restored by imatinib (35), the most widely used tyrosine kinase inhibitor in CML. Another mechanism could be through GRB2-associated–binding protein 2 (GAB2). It has been demonstrated both in vitro, and recently in murine-knockout models, that BCR-ABL1 leukemogenesis is dependent on GAB2 (36, 37). Germline GAB2 mutations have further been linked to childhood leukemia development (38). Meanwhile, Helicobacter pylori uses its virulence factor cytotoxin-associated gene A (CagA) to mimic GAB2 function in the cells it infects (39, 40). CagA is translocated into the cytoplasm of the host cell by a type IV secretion system (41). CagA is then activated in a two-step process, first in a self-limiting step by Src, and then in sustained infections by Abl (42). The tyrosine kinase inhibitor dasatinib, approved for CML, inhibits this phosphorylation in vitro (43). Hence, Helicobacter pylori infections would stand to gain from an increased Abl activity (such as in CML) for CagA phosphorylation, and it could also possibly contribute to BCR-ABL1 leukemogenesis by mimicking GAB2 or by suppressing miR-203.

There are possible confounding factors, which we had no means to control for. Obesity, but not overweight, has in a large meta-analysis been linked to CML risk (44), and could possibly be linked to gastric ulcers (45). However, the risk increase for CML with obesity is modest (RR 1.14 compared with normal weight), and overweight is primarily linked to gastroesophageal reflux and not gastritis or peptic ulcers. The relationship between smoking and CML risk is more controversial, as most studies have not found an association (46), but could possibly have an effect on our results.

Strengths of this study include the population-based approach with clinical data on our CML cases, the size of the CML cohort and the fact that we, compared with the previous study (6), had access to more proxy variables for Helicobacter exposure and a more distinct a priori hypothesis. The obvious limitation is the lack of direct measurements of Helicobacter infection, as our analyses relied on surrogate markers. The backwards truncation of register data further adds to the non-differential misclassification of exposure, where Helicobacter exposure by our markers is likely to be underestimated for both cases and controls. Future studies on CML incidence in existing large-scale cohorts with data on Helicobacter infection status, or on CML cases and healthy controls where convincing biological markers of previous Helicobacter exposure could be measured, would be needed to confirm our findings.

In conclusion, we have demonstrated an increased incidence of CML in patients with previous records of gastritis, dyspepsia, ulcers of peptic and gastric origin as well as among users of PPIs, which are all known proxies for Helicobacter pylori infections. Our data suggest a role for Helicobacter pylori in CML etiology. If the results are substantiated, further research should include whether certain strains of Helicobacter increase risk more than others (which is the case in gastric cancer; ref. 47), and whether eradication of Helicobacter pylori can improve CML prognosis.

No potential conflicts of interest were disclosed.

Conception and design: G. Larfors, L. Stenke, M. Höglund

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Stenke, M. Höglund

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Larfors, J. Richter, A. Själander, L. Stenke, M. Höglund

Writing, review, and/or revision of the manuscript: G. Larfors, J. Richter, A. Själander, L. Stenke, M. Höglund

The authors would like to thank all members of the Swedish CML registry group, the data managers at the Regional Cancer Centers, as well as the Swedish hematologists who have carefully reported patients to the CML register. The work was supported by a generous grant from the Nordic CML Study Group (www.ncmlsg.org).

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.

1.
Hirasuna
JD
,
Shelub
I
,
Bolt
RJ
. 
Hyperhistaminemia and peptic ulcer
.
West J Med
1979
;
131
:
140
3
.
2.
Savage
DG
,
Szydlo
RM
,
Goldman
JM
. 
Clinical features at diagnosis in 430 patients with chronic myeloid leukemia seen at a referral center over a 16-year period
.
Br J Haematol
1997
;
96
:
111
6
.
3.
Karaoglu
AO
,
Kadikoylu
G
,
Yukselen
V
,
Yasa
MH
,
Bolaman
Z
. 
Gastrointestinal lesions and Helicobacter pylori in patients with myeloproliferative disorders
.
Saudi Med J
2004
;
25
:
1913
6
.
4.
Soylu
AR
,
Buyukasik
Y
,
Cetiner
D
,
Buyukasik
NS
,
Koca
E
,
Haznedaroglu
IC
, et al
Overt gastrointestinal bleeding in haematologic neoplasms
.
Dig Liver Dis
2005
;
37
:
917
22
.
5.
Olinger
EJ
,
McCarthy
DM
,
Young
RC
,
Gardner
JD
. 
Hyperhistaminemia and hyperchlorhydria in basophilic granulocytic leukemia
.
Gastroenterology
1976
;
71
:
667
9
.
6.
Johnson
KJ
,
Blair
CM
,
Fink
JM
,
Cerhan
JR
,
Roesler
MA
,
Hirsch
BA
, et al
Medical conditions and risk of adult myeloid leukemia
.
Cancer Causes Control
2012
;
23
:
1083
9
.
7.
Bessede
E
,
Dubus
P
,
Megraud
F
,
Varon
C
. 
Helicobacter pylori infection and stem cells at the origin of gastric cancer
.
Oncogene
2015
;
34
:
2547
55
.
8.
Witkowska
M
,
Smolewski
P
. 
Helicobacter pylori infection, chronic inflammation, and genomic transformations in gastric MALT lymphoma
.
Mediators Inflamm
2013
;
2013
:
523170
.
9.
Houghton
J
,
Stoicov
C
,
Nomura
S
,
Rogers
AB
,
Carlson
J
,
Li
H
, et al
Gastric cancer originating from bone marrow-derived cells
.
Science
2004
;
306
:
1568
71
.
10.
Höglund
M
,
Sandin
F
,
Hellstrom
K
,
Björeman
M
,
Björkholm
M
,
Brune
M
, et al
Tyrosine kinase inhibitor usage, treatment outcome, and prognostic scores in CML: report from the population-based Swedish CML registry
.
Blood
2013
;
122
:
1284
92
.
11.
Sokal
JE
,
Cox
EB
,
Baccarani
M
,
Tura
S
,
Gomez
GA
,
Robertson
JE
, et al
Prognostic discrimination in "good-risk" chronic granulocytic leukemia
.
Blood
1984
;
63
:
789
99
.
12.
Hasford
J
,
Pfirrmann
M
,
Hehlmann
R
,
Allan
NC
,
Baccarani
M
,
Kluin-Nelemans
JC
, et al
A new prognostic score for survival of patients with chronic myeloid leukemia treated with interferon alfa. Writing Committee for the Collaborative CML Prognostic Factors Project Group
.
J Natl Cancer Inst
1998
;
90
:
850
8
.
13.
Wettermark
B
,
Hammar
N
,
Fored
CM
,
Leimanis
A
,
Otterblad Olausson
P
,
Bergman
U
, et al
The new Swedish Prescribed Drug Register—opportunities for pharmacoepidemiological research and experience from the first six months
.
Pharmacoepidemiol Drug Saf
2007
;
16
:
726
35
.
14.
Ludvigsson
JF
,
Andersson
E
,
Ekbom
A
,
Feychting
M
,
Kim
JL
,
Reuterwall
C
, et al
External review and validation of the Swedish National Inpatient Register
.
BMC Public Health
2011
;
11
:
450
.
15.
Charlson
ME
,
Pompei
P
,
Ales
KL
,
MacKenzie
CR
. 
A new method of classifying prognostic comorbidity in longitudinal studies: development and validation
.
J Chronic Dis
1987
;
40
:
373
83
.
16.
Statistics Sweden. Background Facts 2016
:
1, integrated database for labour market research
.
Örebro, Sweden
:
Statistics Sweden
; 
2016
.
17.
Spiegelman
D
,
Hertzmark
E
. 
Easy SAS calculations for risk or prevalence ratios and differences
.
Am J Epidemiol
2005
;
162
:
199
200
.
18.
Berkson
J
,
Gage
RP
. 
Calculation of survival rates for cancer
.
Proc Staff Meet Mayo Clin
1950
;
25
:
270
86
.
19.
Cox
DR
. 
Regression models and life-tables
.
J R Stat Soc Series B Methodol
1972
;
34
:
187
220
.
20.
Sugano
K
,
Tack
J
,
Kuipers
EJ
,
Graham
DY
,
El-Omar
EM
,
Miura
S
, et al
Kyoto global consensus report on Helicobacter pylori gastritis
.
Gut
2015
;
64
:
1353
67
.
21.
Rokkas
T
,
Simsek
I
,
Ladas
S
. 
Helicobacter pylori and non-malignant diseases
.
Helicobacter
2007
;
12
:
20
2
.
22.
Anderson
LA
,
Pfeiffer
RM
,
Landgren
O
,
Gadalla
S
,
Berndt
SI
,
Engels
EA
. 
Risks of myeloid malignancies in patients with autoimmune conditions
.
Br J Cancer
2009
;
100
:
822
8
.
23.
Gunnarsson
N
,
Hoglund
M
,
Stenke
L
,
Wallberg-Jonsson
S
,
Sandin
F
,
Bjorkholm
M
, et al
Increased prevalence of prior malignancies and autoimmune diseases in patients diagnosed with chronic myeloid leukemia
.
Leukemia
2016
;
30
:
1562
7
.
24.
Gunnarsson
N
,
Stenke
L
,
Hoglund
M
,
Sandin
F
,
Bjorkholm
M
,
Dreimane
A
, et al
Second malignancies following treatment of chronic myeloid leukemia in the tyrosine kinase inhibitor era
.
Br J Haematol
2015
;
169
:
683
8
.
25.
Greaves
M
. 
Infection, immune responses and the etiology of childhood leukemia
.
Nat Rev Cancer
2006
;
6
:
193
203
.
26.
Kinlen
L
. 
Childhood leukemia, nuclear sites, and population mixing
.
Br J Cancer
2011
;
104
:
12
8
.
27.
zur Hausen
H
. 
Childhood leukemias and other hematopoietic malignancies: interdependence between an infectious event and chromosomal modifications
.
Int J Cancer
2009
;
125
:
1764
70
.
28.
Molyneux
EM
,
Rochford
R
,
Griffin
B
,
Newton
R
,
Jackson
G
,
Menon
G
, et al
Burkitt's lymphoma
.
Lancet
2012
;
379
:
1234
44
.
29.
Wiemels
JL
,
Talback
M
,
Francis
S
,
Feychting
M
. 
Early infection with cytomegalovirus and risk of childhood hematologic malignancies
.
Cancer Epidemiol Biomarkers Prev
2019
;
28
:
1024
7
.
30.
Francis
SS
,
Wallace
AD
,
Wendt
GA
,
Li
L
,
Liu
F
,
Riley
LW
, et al
In utero cytomegalovirus infection and development of childhood acute lymphoblastic leukemia
.
Blood
2017
;
129
:
1680
4
.
31.
Du
MQ
. 
MALT lymphoma: a paradigm of NF-kappaB dysregulation
.
Semin Cancer Biol
2016
;
39
:
49
60
.
32.
Zhou
X
,
Xu
G
,
Yin
C
,
Jin
W
,
Zhang
G
. 
Down-regulation of miR-203 induced by Helicobacter pylori infection promotes the proliferation and invasion of gastric cancer by targeting CASK
.
Oncotarget
2014
;
5
:
11631
40
.
33.
Bueno
MJ
,
Perez de Castro
I
,
Gomez de Cedron
M
,
Santos
J
,
Calin
GA
,
Cigudosa
JC
, et al
Genetic and epigenetic silencing of microRNA-203 enhances ABL1 and BCR-ABL1 oncogene expression
.
Cancer Cell
2008
;
13
:
496
506
.
34.
Craig
VJ
,
Cogliatti
SB
,
Rehrauer
H
,
Wundisch
T
,
Muller
A
. 
Epigenetic silencing of microRNA-203 dysregulates ABL1 expression and drives Helicobacter-associated gastric lymphomagenesis
.
Cancer Res
2011
;
71
:
3616
24
.
35.
Shibuta
T
,
Honda
E
,
Shiotsu
H
,
Tanaka
Y
,
Vellasamy
S
,
Shiratsuchi
M
, et al
Imatinib induces demethylation of miR-203 gene: an epigenetic mechanism of anti-tumor effect of imatinib
.
Leuk Res
2013
;
37
:
1278
86
.
36.
Gu
S
,
Chan
WW
,
Mohi
G
,
Rosenbaum
J
,
Sayad
A
,
Lu
Z
, et al
Distinct GAB2 signaling pathways are essential for myeloid and lymphoid transformation and leukemogenesis by BCR-ABL1
.
Blood
2016
;
127
:
1803
13
.
37.
Halbach
S
,
Kohler
M
,
Uhl
FM
,
Huber
J
,
Zeiser
R
,
Koschmieder
S
, et al
Gab2 is essential for Bcr-Abl-mediated leukemic transformation and hydronephrosis in a chronic myeloid leukemia mouse model
.
Leukemia
2016
;
30
:
1942
5
.
38.
de Smith
AJ
,
Lavoie
G
,
Walsh
KM
,
Aujla
S
,
Evans
E
,
Hansen
HM
, et al
Predisposing germline mutations in high hyperdiploid acute lymphoblastic leukemia in children
.
Genes Chromosomes Cancer
2019
;
58
:
723
30
.
39.
Simister
PC
,
Feller
SM
. 
Order and disorder in large multi-site docking proteins of the Gab family–implications for signalling complex formation and inhibitor design strategies
.
Mol Biosyst
2012
;
8
:
33
46
.
40.
Botham
CM
,
Wandler
AM
,
Guillemin
K
. 
A transgenic Drosophila model demonstrates that the Helicobacter pylori CagA protein functions as a eukaryotic Gab adaptor
.
PLoS Pathog
2008
;
4
:
e1000064
.
41.
Yamaoka
Y
,
Graham
DY
. 
Helicobacter pylori virulence and cancer pathogenesis
.
Future Oncol
2014
;
10
:
1487
500
.
42.
Tammer
I
,
Brandt
S
,
Hartig
R
,
Konig
W
,
Backert
S
. 
Activation of Abl by Helicobacter pylori: a novel kinase for CagA and crucial mediator of host cell scattering
.
Gastroenterology
2007
;
132
:
1309
19
.
43.
Krisch
LM
,
Posselt
G
,
Hammerl
P
,
Wessler
S
. 
CagA phosphorylation in Helicobacter pylori-infected B cells is mediated by the nonreceptor tyrosine kinases of the Src and Abl families
.
Infect Immun
2016
;
84
:
2671
80
.
44.
Castillo
JJ
,
Reagan
JL
,
Ingham
RR
,
Furman
M
,
Dalia
S
,
Merhi
B
, et al
Obesity but not overweight increases the incidence and mortality of leukemia in adults: a meta-analysis of prospective cohort studies
.
Leuk Res
2012
;
36
:
868
75
.
45.
Sogabe
M
,
Okahisa
T
,
Kimura
T
,
Okamoto
K
,
Miyamoto
H
,
Muguruma
N
, et al
Influence of metabolic syndrome on upper gastrointestinal disease
.
Clin J Gastroenterol
2016
;
9
:
191
202
.
46.
Kabat
GC
,
Wu
JW
,
Moore
SC
,
Morton
LM
,
Park
Y
,
Hollenbeck
AR
, et al
Lifestyle and dietary factors in relation to risk of chronic myeloid leukemia in the NIH-AARP Diet and Health Study
.
Cancer Epidemiol Biomarkers Prev
2013
;
22
:
848
54
.
47.
McClain
SM
,
Beckett
CA
,
Cover
LT
. 
Helicobacter pylori vacuolating toxin and gastric cancer
.
Toxins (Basel)
2017
;
9
.
doi: 10.3390/toxins9100316
.