Association Between Proton Pump Inhibitors and the Risk of Hepatocellular Carcinoma

Y.-H. J. Shao; T.-S. Chan; K. Tsai; S.-Y. Wu

Aliment Pharmacol Ther. 2018;48(4):460-468.

Abstract and Introduction

Abstract

Background: Proton pump inhibitor (PPI) use has been reported to be associated with liver damage and might possibly be carcinogenic.

Aims: We examined whether long–term PPI use increases the risk of hepatocellular carcinoma (HCC) in patients without viral hepatitis B or C.

Methods: We conducted a nested case–control study in a cohort of patients without viral hepatitis in Taiwan from 2000 to 2013. In total, 29 473 HCC cases and 294 508 matched controls were included. Moreover, we identified prescriptions for PPI and durations between the PPI index date and cancer diagnosis date (or the corresponding date in controls).

Results: The adjusted odds ratio (AOR) for HCC associated with PPI use was 2.86 (95% confidence interval [CI], 2.69–3.04). Considering the use of PPIs determined according to cumulative defined daily dose (cDDD) subgroups, a dose–response effect was observed in patients exposed to 29–180, 181–240, 241–300, and 300+ cDDDs of PPIs. The AORs were 2.74 (95% CI, 2.57–2.93), 2.98 (95% CI, 2.50–3.56), 3.23 (95% CI, 2.59–4.02), and 3.43 (95% CI, 2.94–4.00) in the 29–180, 181–240, 241–300, and 300+ cDDD groups, respectively, compared with the 0–28 cDDD group. A sensitivity analysis revealed a consistent association between PPI use and the risk of HCC in subpopulations stratified by risk factors associated with HCC.

Conclusions: This observational study demonstrated that PPIs might increase the risk of HCC.

Introduction

Proton pump inhibitors (PPIs) are the most potent inhibitors of gastric acid secretion.^[1,2] H–K–ATPase constitutes the final pathway through which HCl is secreted into the gastric lumen, where it hydrolyses dietary proteins as well as maintains a sterile environment.^[1] PPIs inhibit H–K–ATPase^[1] and substantially reduce acid secretion with a long duration of action.^[3,4] Consequently, PPIs are commonly prescribed for patients with gastro–oesophageal reflux disease (GERD), peptic oesophagitis and oesophagitis–associated symptoms.^[3,4]

One of the clinical consequences of strong and long–term acid suppression is hypergastrinemia. The resultant hypergastrinemia increases the risk of cancer development.^[5] First, hyperplasia of enterochromaffin–like cells (ECL cells) was observed in rodents receiving PPIs,^[6] leading to gastric carcinoid tumours. Although gastric carcinoid tumours have not been detected in humans since the introduction of PPIs, hyperplasia of ECL cells was observed in clinical studies.^[7] Moreover, fundic gland polyposis was observed in patients receiving potent and prolonged antisecretory therapy.^[8] In addition to cancer observed in the stomach, hypergastrinemia in PPI users was associated with pancreatic cancer,^[9] periampullary cancers,^[5] colorectal cancer,^[10] lung cancer^[11] and ovarian cancer.^[12]

Reports on the hepatotoxicity of PPIs are scarce in the literature.^[13–15] However, a recent study used genome–wide transcriptional expression data from cultured human cells treated with bioactive molecules to study the carcinogenic effect of compounds on the human liver.^[16] A PPI was reported to exhibit a notable genetic expression similar to well–known carcinogens in the liver;^[16,17] however, its carcinogenicity remains unknown. Moreover, PPIs allow the proliferation of cells with fatal mutations, and they thus increase the risks of cancers.^[18] The mechanism is probably indirect and involves the induction of oxidative stress and production of reactive oxygen species that further damage DNA. Repeated DNA damage likely increases the mutation rate, including that of tumour suppressor genes and oncogenes.^[19] The mutation burden engendered by dose–dependent PPI use in tumour suppressor genes and oncogenes might affect the onset of hepatocellular carcinoma (HCC).^[20,21] In addition to the reported association between HCC risk and PPI use at the molecular level, we hypothesised that PPI use increases the risk of HCC. In this study, we examined the association between PPI use and the risk of HCC and the impact of long–term PPI use.

1 2 3 4 5

Next Section

Table 1. Characteristics of patients with hepatocellular carcinoma and their matched controls
Variables	HCC cases (n = 29 473)		Matched controls (n = 294 508)		P value
Sex, n, %					0.9738
Male	20 083	68.1%	200 709	68.2%
Female	9390	31.9%	93 799	31.8%
Age at index date, n, %
<40	1272	4.30%	12 336	4.20%	0.4609
40–59	9259	31.42%	93 280	31.67%
60–79	14 384	48.80%	143 763	48.81%
80+	4558	15.50%	45 129	15.30%
Mean follow–up duration in mo (SD)	60.1 (48.35)		60.05 (48.32)		0.9072
Year of index date, n, %					0.9943
2002–2005	8771	29.8%	87 692	29.8%
2006–2009	8035	27.3%	80 337	27.3%
2010–2013	12 667	43.0%	126 479	42.9%
PPI use >28 cDDD					<0.0001
Yes	2448	8.31%	7840	2.66%
No	27 025	91.69%	286 668	97.34%
Coexisting medical condition, n, %
Hypertension	6708	22.8%	55 207	18.7%	<0.0001
Diabetes	3619	12.3%	18 182	6.2%	<0.0001
Chronic obstructive pulmonary disease	2665	9.0%	22 819	7.7%	<0.0001
Acute coronary syndrome	2289	7.8%	18 988	6.4%	<0.0001
Cerebrovascular accident	1611	5.5%	11 417	3.9%	<0.0001
Peptic ulcer disease	2351	8.0%	12 402	4.2%	<0.0001
Gastro–oesophageal reflux disease	2973	10.1%	15 152	5.1%	<0.0001
Cirrhosis	1024	3.5%	1109	0.4%	<0.0001
Hyperlipidaemia	1846	6.3%	15 314	5.2%	<0.0001
Statins (% within hyperlipidaemia patients)	823	44.6%	5503	35.9%	<0.0001
Metformin (% within DM patients)	2123	58.7%	10 131	55.7%	0.0011
H. pylori eradication therapy	868	2.9%	2673	0.9%	<0.0001
H2–receptor antagonists	4921	16.7%	33 509	11.4%	<0.0001
Aspirin	4360	14.8%	36 009	12.2%	<0.0001
NSAIDs	7558	25.6%	82 934	28.2%	<0.0001
Charlson comorbidity index					<0.0001
0	19 579	66.4%	236 864	80.4%
1	4931	16.7%	41 328	14.0%
2	2935	10.0%	11 932	4.1%
≥3	2028	6.9%	4384	1.5%

PPI, proton pump inhibitor; cDDD, cumulative defined daily dose; DM, diabetes mellitus; HCC, hepatocellular carcinoma.

Table 2. Crude and adjusted odds ratio of hepatocellular carcinoma risk associated with proton pump inhibitor use and other covariates
Variables	OR (95% CI)	aOR (95% CI)^a
PPI use >28 cDDD vs not	4.30 (4.07–4.54)	2.86 (2.69–3.04)
Previous or coexisting medical condition
Hypertension	1.31 (1.27–1.35)	1.06 (1.03–1.10)
Diabetes	2.18 (2.10–2.26)	1.99 (1.91–2.08)
Chronic obstructive pulmonary disease	1.19 (1.14–1.24)	1.07 (1.02–1.12)
Acute coronary syndrome	1.23 (1.18–1.29)	0.98 (0.93–1.04)
Cerebrovascular accident	1.45 (1.37–1.53)	1.19 (1.12–1.26)
Peptic ulcer disease	2.00 (1.91–2.09)	1.35 (1.28–1.43)
Gastro–oesophageal reflux disease	2.10 (2.02–2.19)	3.65 (3.15–4.22)
Cirrhosis	9.61 (8.82–10.48)	7.88 (7.20–8.63)
Hyperlipidaemia	1.22 (1.16–1.29)	0.85 (0.80–0.90)
H. pylori eradication therapy	3.41 (3.15–3.69)	1.78 (1.63–1.95)
H2–receptor antagonists	1.57 (1.52–1.62)	1.41 (1.36–1.46)
Aspirin	1.26 (1.22–1.30)	1.05 (1.01–1.10)
NSAIDs	0.88 (0.85–0.90)	0.74 (0.72–0.76)

cDDD, cumulative defined daily dose; OR, odds ratio; 95% CI, 95% confidence interval; NSAIDs, nonsteroidal anti–inflammatory drugs.
^aAdjusted odds ratio was estimated using conditional logistic regression adjusted for other covariates listed in the table.

Table 3. Distribution of proton pump inhibitor use among hepatocellular carcinoma cases and their matched controls
PPI use	HCC cases (n = 29 473)	Matched controls (n = 294 508)	OR (95% CI)	aOR (95% CI)^a	P value (trend test)
PPI use (cDDD >28)	2448	7840	4.30 (4.07–4.54)	2.86 (2.69–3.04)	—
Cumulative dose					<0.0001
0 < cDDD ≤ 28	27 133	287 134	1.00	—
28 < cDDD ≤ 180	1722	5777	4.05 (3.81–4.30)	2.74 (2.57–2.93)
180 < cDDD ≤ 240	194	578	4.57 (3.87–5.40)	2.98 (2.50–3.56)
240 < cDDD ≤ 300	129	337	5.10 (4.15–6.27)	3.23 (2.59–4.02)
300 < cDDD	295	682	5.92 (5.14–6.82)	3.43 (2.94–4.00)

The Cochran–Armitage trend test was used to test whether adverse responses increase with larger drug doses.
cDDD, cumulative defined daily dose; OR, odds ratio; 95% CI, 95% confidence interval; PPI, proton pump inhibitor; HCC, hepatocellular carcinoma.
^aAdjusted odds ratio was estimated from multivariate conditional logistic regression. Other covariates included in the model were hypertension, diabetes, chronic obstructive pulmonary disease, acute coronary syndrome, cerebrovascular accident, peptic ulcer disease, gastro–oesophageal reflux disease, cirrhosis and hyperlipidaemia.

Table 4. Adjusted odds ratio of hepatocellular carcinoma risk associated with proton pump inhibitor use in subpopulation analysis
Subpopulation	Cases	Controls	Crude odds ratio (95% CI)		Adjusted odds ratio (95% CI)^a
Cirrhosis matched
Cohort with cirrhosis
PPI user	1222	4323	4.04	(3.74–4.38)	2.85	(2.62–3.10)
Non–PPI user	11 526	122 720	1	—	1	—
Cohort without cirrhosis
PPI user	1224	15 490	3.48	(3.23–3.75)	2.33	(2.15–2.29)
Non–PPI user	4552	162 382	1	—	1	—
Diabetes matched
Cohort with diabetes^b
PPI user	1063	3718	4.05	(3.72–4.40)	2.91	(2.65–3.19)
Non–PPI user	8513	91 896	1	—	1	—
Cohort without diabetes
PPI user	1308	4083	4.31	(3.99–4.62)	2.85	(2.63–3.09)
Non–PPI user	18 571	194 441	1	—	1	—
Hyperlipidaemia matched
Cohort with hyperlipidaemia^c
PPI user	884	2811	4.64	(4.22–5.10)	2.93	(2.64–3.26)
Non–PPI user	6066	66 510	1	—	1	—
Cohort without hyperlipidaemia
PPI user	1495	4374	4.54	(4.24–4.86)	2.86	(2.65–3.08)
Non–PPI user	20 653	215 058	1	—	1	—
H. pylori eradication therapy matched
Cohort with treatment
PPI user	1169	5789	3.08	(2.83–3.36)	2.07	(1.89–2.27)
Non–PPI user	3853	44 066	1	—	1	—
Cohort without treatment
PPI user	1298	5181	3.05	(2.85–3.27)	2.37	(2.17–2.52)
Non–PPI user	23 429	241 871	1	—	1	—
Gastro–oesophageal reflux disease
Cohort with GERD
PPI user	955	5881	2.05	(1.88–2.24)	1.05	(0.94–1.16)
Non–PPI user	3808	41 187	1	—	1	—
Cohort without GERD
PPI user	1510	4700	4.24	(3.96–4.54)	2.97	(2.76–3.20)
Non–PPI user	23 474	244 952	1	—	1	—

The base model included cirrhosis, hypertension, diabetes, chronic obstructive pulmonary disease, acute coronary syndrome, cerebrovascular accident, peptic ulcer disease, gastro–oesophageal reflux disease, hyperlipidaemia, H. pylori eradication therapy, H2–receptor antagonists, aspirin and NSAIDs. In each condition–matched population, the variable being matched was excluded from the base model.
PPI, proton pump inhibitor; GERD, gastro–oesophageal reflux disease; NSAIDs, nonsteroidal anti–inflammatory drugs; 95% CI, 95% confidence interval.
^aAdjusted odds ratio was estimated using multivariate conditional logistic regression.
^bMetformin was included in the adjustment.
^cStatins were included in the adjustment.

Supplemental Table 1. Sensitivity analysis estimating the effect of an unmeasured confounder on the odds ratio of hepatocellular carcinoma.
Prevalence in patients with PPI, %	Prevalence in patients without PPI, %	Risk of HCC adjusted for unmeasured confounder (95% CI)
Prevalence in patients with PPI, %	Prevalence in patients without PPI, %	Unmeasured confounder OR 1.30	Unmeasured confounder OR 2.00	Unmeasured confounder OR 2.50
10	60	2.35 (2.21, 2.50)	1.63 (1.53, 1.73)	1.31 (1.23, 1.39)
	40	2.28 (2.33, 2.64)	1.86 (1.75, 1.97)	1.55 (1.46, 1.65)
	20	2.62 (2.46, 2.78)	2.17 (2.04, 2.30)	1.91 (1.80, 2.03)
20	60	2.29 (2.15, 2.43)	1.49 (1.40, 1.58)	1.16 (1.09, 1.23)
	40	2.41 (2.27, 2.56)	1.70 (1.60, 1.81)	1.38 (1.29, 1.46)
	20	2.55 (2.39, 2.71)	1.99 (1.87, 2.11)	1.69 (1.59, 1.80)
30	60	2.22 (2.09, 2.36)	1.38 (1.29, 1.46)	1.04 (0.98, 1.10)
	40	2.34 (2.20, 2.49)	1.57 (1.48, 1.67)	1.23 (1.16, 1.31)
	20	2.48 (2.33, 2.63)	1.83 (1.72, 1.95)	1.52 (1.43, 1.61)

HCC, hepatocellular carcinoma; PPI, proton pump inhibitor; 95% CI, 95% confidence interval; OR, odds ratio.

Supplemental Table 2. PPIs and risks of kidney cancer and head and neck cancer.
Cancer type	No of Cases	No of Controls	Adjusted Odds Ratio (95% CI)
Kidney cancer	32,957	329,408	0.99 (0.97–1.03)
Head and neck cancer	12,464	123,476	0.93 (0.82–1.01)

* Adjusted odds ratio was estimated using multivariate conditional logistic regression.
The base model included cirrhosis, hypertension, diabetes, chronic obstructive pulmonary disease, acute coronary syndrome, cerebrovascular accident, peptic ulcer disease, gastroesophageal reflux disease, hyperlipidemia, H. pylori eradication therapy, H2-receptor antagonists, aspirin, and NSAIDs.