Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Long-term exposure to antihypertensive drugs and
the risk of cancer occurrence: evidence from a large
population-based study
Matteo Franchi
a,b
, Ginevra Torrigiani
a,b
, Sverre Erik Kjeldsen
c
, Giuseppe Mancia
d
,
and Giovanni Corrao
a,d
Objective: Available data on the association between
antihypertensive drugs and cancer are characterized by a
few years follow-up. Our aim has been to evaluate the
association between long-term exposure to
antihypertensive drugs and the risk of cancer occurrence.
Methods: Using the healthcare utilization databases of the
Lombardy region (Italy), individuals aged 40– 85 years who
had no previous history of cancer and were newly
dispensed with at least one antihypertensive drug from the
major drug classes between 2009 and 2011 were followed
from the first drug dispensation to December 31, 2020.
Data were analyzed according to the first drug used and
the intention to treat principle, but also via an ‘‘as
treated’’ approach, that is, by considering changes of and
exposure to drugs during follow-up. The association
between the duration of exposure to each drug class and
the risk of cancer occurrence was evaluated using the
adjusted Cox regression models.
Results: The study cohort included 338 910 new drug
users (median age, 59 years; 49.5% males). During a
median follow-up of 10.2 years, 36 556 cancers occurred.
There was no consistent significant association between
the risk of cancer occurrence and angiotensin-converting-
enzyme inhibitors, angiotensin-receptor blockers, or
thiazides. A progressive, weak increase in cancer
occurrence was associated with progressive exposure to
calcium channel blockers and, limited to long-term
exposure, to beta-blockers. A modest progressive increase
in risk was observed also for thiazide-like and loop
diuretics in the as treated, although not in the intention to
treat approach.
Conclusions: Long-term evaluation of exposure to
antihypertensive drugs did not show consistent
associations between thiazides, angiotensin-receptor
blockers, or angiotensin-converting-enzyme inhibitors and
the risk of cancer occurrence. A weak association was
observed between cancer and the duration of exposure to
calcium channel blockers and beta-blockers.
Keywords: antihypertensive drugs, cancer, long-term
follow-up, observational study, population-based
Abbreviations: ACEIs, angiotensin converting enzyme
inhibitors; ARBs, angiotensin receptor blockers; ATC,
anatomical therapeutic chemical; BBs, beta blockers; CCBs,
calcium channel blockers; CI, confidence intervals; Ds,
diuretics; HR, hazard ratios; ICD-9-CM, International
Classification of Diseases, ninth revision, Clinical
Modification; Loop D, loop diuretics; MCS, Multisource
Comorbidity Score; RHS, Regional Health Service
BACKGROUND
The association between antihypertensive drugs and
the risk of cancer has been investigated in both
observational studies andrandomizedtreatment
trials. Observational studies and their meta-analyses have
provided variable results. Namely, while frequently con-
cluding that exposure to antihypertensive drugs does not
substantially modify the risk of cancer, these studies have
also reported modest reductions and increases either in
the overall cancer risk or in site-specific cancers [1–13],
such in the case of thiazides and the increased risk of skin
cancer [14–16]. In contrast, meta-analyses of randomized
trials have usually concluded, in the context of a better
comparability of antihypertensive drug users and con-
trols, that major antihypertensive drug classes (diuretics,
beta-blockers, angiotensin converting enzyme(ACE)
inhibitors, angiotensin receptor antagonists, and calcium
channel blockers) are associated with no or only a very
modest change in the risk of cancer [17–22]. However,
available data on the association between antihyperten-
sive drugs and cancer risk have an important limitation,
Journal of Hypertension 2024, 42:2107–2114
a
National Centre for Healthcare Research and Pharmacoepidemiology,
b
Unit of
Biostatistics, Epidemiology and Public Health, Department of Statistics and Quantita-
tive Methods, University of Milano-Bicocca, Milan, Italy,
c
Institute for Clinical Medi-
cine, University of Oslo, Oslo, Norway and
d
University of Milano-Bicocca, Milan, Italy
Correspondence to Dr Matteo Franchi, Dipartimento di Statistica e Metodi Quanti-
tativi, Universit
a degli Studi di Milano-Bicocca, Via Bicocca degli Arcimboldi, 8, Edificio
U7, 20126 Milano, Italy. Tel: +39 02 64485832; e-mail: matteo.franchi@unimib.it
Received 22 July 2023 Revised 13 July 2024 Accepted 4 August 2024
J Hypertens 42:2107– 2114 Copyright ©2024 The Author(s). Published by Wolters
Kluwer Health, Inc. This is an open access article distributed under the terms of the
Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-
NC-ND), where it is permissible to download and share the work provided it is properly
cited. The work cannot be changed in any way or used commercially without
permission from the journal.
DOI:10.1097/HJH.0000000000003841
Journal of Hypertension www.jhypertension.com 2107
Original Article
that is, the short duration of the treatment period compared
with the exposure necessary for many well established
carcinogenic factors to make their effect clinically manifest
[23]. This limitation is particularly evident for randomized
trials which, in the most recent meta-analyses, have reported
an average follow-up only slightly longer than 4 years [17].
The short duration of drug exposure is an even greater
problem if one considers that available studies do not take
into account that adherence to antihypertensive drug treat-
ment is usually suboptimal and variable between patients
[24,25], making true drug exposure to the possible carcino-
genic effect of antihypertensive drugs less than that reported
by the available studies and different between patients to an
unmeasured degree.
The aim of our study was to use a large real-life database of
an Italian population to analyze the relationship between
antihypertensive drugs and the risk of cancer occurrence on
a much longer follow-up than that of randomized trials and
most observational studies. Another new aspect of the study
was that exposure to antihypertensive drugs was measured
according to the availability of antihypertensive drug pre-
scriptions during the entire follow-up period, thus more
precisely assessing patients’exposure to any given BP-low-
ering agent [26]. Data were analyzed for the major antihyper-
tensive agents recommended by hypertension guidelines
[27], i.e. diuretics (Ds), angiotensin converting enzyme inhib-
itors (ACEIs), angiotensin receptor blockers (ARBs), calcium
channel blockers (CCBs) and beta blockers (BBs).
METHODS
Data sources
In this retrospective cohort study, we used the popula-
tion-based healthcare utilization databases of Lombardy,
the largest Italian Region, and the one which accounts for
approximately 10 million inhabitants. Data extend to all
the beneficiaries of the Regional Health Service (RHS)
(almost the entire Lombardy population), and include
demographic information (sex, year of birth, dates of start
and end of the condition of RHS beneficiary, vital status,
and date of death) and healthcare-related data, including
diagnoses and interventions [coded through the Interna-
tional Classification of Diseases, ninth revision, Clinical
Modification (ICD-9-CM) – classification system] by all
hospitals of the region, drugs prescribed by the RHS
physicians to outpatients, directly dispensed by territorial
pharmacies [coded through the anatomical therapeutic
chemical (ATC) – classification system] and completely or
almost completely reimbursed by the RHS, and co-pay-
ment exemptions per disease. Antihypertensive drugs (as
well as other lifesaving drugs) are also given free or almost
free of charge by the RHS to all citizens with no temporal
treatment limits, in line with the therapeutic principle to
treat hypertension for the lifetime. A unique anonymized
individual identification code was used to link each data-
base to the others. Details of the regional databases and
their use for pharmacoepidemiology studies in the setting
of hypertension are reported elsewhere [28]. The list of
ICD-9-CMandATCcodesusedinthecurrentstudyare
reported in Table 1, Supplemental Digital Content, http://
links.lww.com/HJH/C544.
Cohort selection
The initial cohort included all beneficiaries of the RHS who
were dispensed with at least one antihypertensive drug
from January 1, 2009, to December 31, 2011. The date of the
first dispensation was labeled as ‘‘index date’’. To identify
new users of antihypertensive drugs, patients with any
prescription of one or more antihypertensive drugs during
the five years preceding the index date were excluded.
Moreover, we excluded patients aged less than 40 or
>85 years at the index date, those who were not benefi-
ciaries of the RHS for at least 5 years before the index date,
and those with either an ICD-9-CM diagnostic code of
malignant cancer, a dispensation of one or more antineo-
plastic drugs, or an exemption from co-payment for drugs
or medical services related to cancer during the five years
preceding the index date. Patients with less than 1 year of
observation after the index date were also excluded. The
remaining patients comprised the final study cohort.
Definition of exposure to antihypertensive
drugs and follow-up
Antihypertensive drugs were classified into the following
classes: ACEIs, ARBs, CCBs, BBs, and Ds. Diuretics were
further classified into thiazides and thiazide-likes (Chlor-
thalidone and Indapamide), and the analysis was extended
to loop diuretics (Loop Ds). The list of drugs included in the
analysis is reported in Table 1, Supplemental Digital Con-
tent, http://links.lww.com/HJH/C544 according to the
drug class.
The exposure of the cohort members to each class of
antihypertensive drug was evaluated using two
approaches. First, using an ‘‘intention-to-treat’’ approach,
each cohort member was included in the class of drugs
received at the index date without considering subsequent
drug discontinuation, adding on or switching from one
drug class to another during the follow-up. Second, using
an ‘‘as-treated’’ approach, the exposure of each cohort
member was evaluated by considering the duration of
the therapeutic pathway, that is, the duration of exposure
to each class of drugs, measured in months and inferred
from a drug prescription covering at least 21 days (or
28 days, see below) of the month. Assuming that no pre-
scription meant no drug availability (an assumption sup-
ported by the evidence that prescriptions issued by
nonreimbursable private medical visits not included in
the databases represent a very small fraction of the overall
antihypertensive drug use) [29], the duration of exposure to
each class of drugs was categorized as: 0 (unexposed), 1 –
24, 25–48, 49–72, 73 –96, or >96 months. The follow-up of
each cohort member started at the index date and ended at
the earliest date between the onset of cancer (the clinical
outcome of interest, see next paragraph), censoring due to
migration to other regions, death, or December 31, 2020.
Outcome definition
The clinical outcome of interest was the occurrence of
malignant cancer, defined as the first ICD-9-CM diagnostic
code for malignant cancer recorded during follow-up. Only
cancers that occurred after one year of follow-up
were considered.
Franchi et al.
2108 www.jhypertension.com Volume 42 Number 12 December 2024
Baseline characteristics
Individual patient characteristics included sex and age at
the index date plus co-treatment with drugs such as statins,
antidiabetic agents, other cardiovascular drugs, and non-
steroidal anti-inflammatory drugs. Database information
also included hospitalization for respiratory, hepatic, renal,
cardiovascular, and other diseases during the two years
before the index date. In each patient calculation was made
of the Multisource Comorbidity Score (MCS), which
includes 34 different clinical conditions weighted for their
association with hospitalization and death and has been
found to accurately predict the risk of hospitalization and
death in the Italian population [30].
Statistical analyses
Baseline characteristics were reported as absolute frequen-
cies and percentages and were calculated both for the
overall cohort and for cohort stratification according to
exposure to each class of drugs for at least one month
during the follow-up.
The association between exposure to the class of drugs
received at the index date (intention-to-treat approach), as
compared to other classes of drugs, and the risk of cancer
occurrence was evaluated by the Cox regression models,
adjusted for sex, age, and MCS classes. This was done also
when the occurrence of cancer was assessed by the ‘‘as
treated’’ approach, in which the duration of exposure to
each class of drugs was included in the model as a time-
dependent covariate. The model was mutually adjusted for
each class of drugs considered. Estimates were reported as
hazard ratios (HRs) along with their 95% confidence inter-
vals (CIs).
The SAS statistical package (SAS Institute Inc., version
9.4, Cary, NC, USA) was used for all statistical analyses.
Sensitivity analyses
To evaluate the robustness of the results four sensitivity
analyses were performed. To account for the effects of
changes in antihypertensive treatment strategy during fol-
low-up, (e.g. switching from initial treatment to another
class of drugs), in the first sensitivity analysis the follow-up
was censored at the date of the change of therapeutic
strategy. To further account for the possibility that the
treatment change may have been related to outcome oc-
currence (generating biased estimate by informative cen-
soring) an inverse probability-of-censoring weights (IPCW)
approach was used [31]. A weight inversely proportional to
the probability of censoring was applied to each observa-
tion, and the weight was quantified using a time-dependent
Cox regression model including the available time-varying
variables measured from index date until the date of the
switching therapy. The time-varying variables included
hospital admissions, outpatient services (visits, diagnostic
procedures, and laboratory tests), and drugs dispensed
from the initial treatment until the treatment switch. Sepa-
rate models were implemented for patients who received
each class of drugs [32]. The censoring weights were stabi-
lized using the probability of censoring, conditional on the
class of drugs received [33]. The stabilized weights were
truncated at 5th and 95th percentile of their distribution in
order to reduce their variability [34], and were used to
estimate the marginal Cox regression model parameters
for assessing the exposure–outcome association using HR
and the corresponding 95% CI. The estimates obtained
using the above procedure were compared with those
obtained from the main analyses.
Second, because the mutual adjustment for each drug
classes as it was done in the main analysis may lead to
mistaken interpretations of these estimates [35], in the
second sensitivity analysis the association between each
drug class and cancer occurrence was separately evaluated
through Cox regression models adjusted for sex, age and
MCS classes only.
In the third sensitivity analysis, the definition of monthly
exposure to a given drug class was based on a prescription
coverage of at least 28 days, instead of 21 days, as in the
main analysis. Finally, in the fourth sensitivity analysis, the
date of cancer occurrence was defined by the correspond-
ing ICD-9-CM diagnostic codes (as in the main analysis) or
by the initial prescription of antineoplastic drugs to account
for cancers that may not have been diagnosed in the
hospital setting or in which an ICD-9-CM diagnostic code
may have not been recorded.
RESULTS
Cohort selection and characteristics
Overall, 2 550 941 users of antihypertensive drugs were
identified between the period 2009 and 2011. Of these, 490
967 were new users. After applying the exclusion criteria,
338 910 individuals were included in the study cohort. A
flowchart of cohort selection is shown in Fig. 1.
In the entire cohort, the median age at the index date was
59 years, and 49.5% of the patients were males. Respiratory,
hepatic, and kidney diseases affected 2.6%, 0.7%, and 0.4%
of cohort members, respectively. Statins, nonsteroid anti-
inflammatory drugs, and antidiabetic drugs were used by
15.6%, 32.6%, and 6.9% of patients, respectively. During
follow-up, the number of individuals exposed for at least
one month to ACEIs, ARBs, thiazides, thiazides-like, Loop
D, BBs, and CCBs were 193 969, 144 176, 128 337, 8616,
39 569, 118 205, and 124 342, respectively. Between the
different classes of drugs, the median age varied between
59 and 68 years, and the percentage of males varied be-
tween 38.8% (patients treated with thiazides-like drugs)
and 53.4% (patients treated with CCBs). The distribution
of MCS, as well as the prevalence of the diseases and co-
treatments, did not vary substantially between different
treatment groups, except for Loop D, which were associat-
ed with a worse clinical profile (Table 1).
Exposure to antihypertensive drugs and risk of
cancer occurrence. Intention-to-treat approach
During a median follow-up of 10.2 years, 36 556 cancers
occurred. The overall number of cancers was 19 967 for
ACEIs, 13 826 for ARBs, 12 190 for thiazides, 712 for
thiazide-like, 4036 for Loop D, 10 928 for BBs, and 11
305 for CCBs. As shown in Fig. 2 the intention-to-treat
approach generated evidence of a modest cancer risk
reduction in patients who started antihypertensive treat-
ment with ACEIs (HR ¼0.96, 95% CI 0.94 –0.98), together
Antihypertensive drugs and cancer risk
Journal of Hypertension www.jhypertension.com 2109
with a borderline risk reduction in those treated with ARBs
(HR ¼0.98, 95% CI 0.95–1.00). The cancer risk was not
modified by thiazides, thiazides-like, or BBs, whereas a
borderline excess was seen in patients initially treated with
CCBs (HR ¼1.03, 95% CI 1.00–1.06) and Loop Ds
(HR ¼1.05, 95% CI 1.00–1.11).
Exposure to antihypertensive drugs and risk of
cancer occurrence. As treated approach
As shown in Fig. 3, when considering exposure to antihy-
pertensive drugs during the entire follow-up (as-treated
approach), no consistent trend of cancer risk modification
with an increasing duration of exposure was observed for
ACEIs, ARBs, and thiazide diuretics. Thiazide-like and loop
diuretics showed a trend toward an increased risk of cancer
as exposure increased, in the context of a relatively small
number of patients and events and wide confidence inter-
vals with longer exposures. An increased risk of cancer was
observed in patients treated with BBs only after 8 years of
exposure (HR ¼1.33, 95% CI 1.10– 1.60) whereas in
patients treated with CCBs, there was a trend toward a
modest progressive increase in cancer risk, especially for
exposure durations longer than 4 years. Similar results were
observed in males and females (Figures S1 and S2, Supple-
mental Digital Content, http://links.lww.com/HJH/C544).
Sensitivity analyses
By adopting the IPCW approach, 20 339 cancers were
identified during a median follow-up of 8.0 years. The results
derived from this analysis confirmed those obtained in the
main analysis (Figure S3, Supplemental Digital Content,
http://links.lww.com/HJH/C544). The main results did not
change substantially when estimates were only adjusted for
sex, age, and MCS classes, except for an increased risk for
Loop Ds, or when we considered a 28-day coverage with a
given class of drugs for defining exposure in a given month of
follow-up. The corresponding results are shown in Figures
S4 and S5, Supplemental Digital Content, http://links.lww.
com/HJH/C544, respectively . The secondary definition of
cancer onset, based on both ICD-9-CM diagnostic codes of
cancer and dispensation of antineoplastic drugs, led to the
identification of 40 096 cancers. The estimates of the associa-
tion between exposure duration and the risk of cancer
occurrence confirmed those generated by the main analysis
(Figure S6, Supplemental Digital Content, http://links.lww.
com/HJH/C544).
New users of antihypertensive drugs
(index date)
(N = 490,035)
Patients included in the final study
cohort
(N = 338,910)
Excluded:
•Age <40 or > 85 years (N = 67,699)
•Beneficiaries of the Regional
Health Service for less than 5 years
from index date (N = 40,030)
•Diagnostic code of malignant
neoplasm, dispensation of
antineoplastic drugs or exemption
for cancer in the 5 years preceding
the index date (N = 33,059)
•Follow-up less than one year (N =
8,691)
•Errors/inconsistent data (N = 1,646)
Beneficiaries of the Regional Health
Service with at least one dispensation
of antihypertensive drug between
January 1, 2009 and December 31,
2011
(N = 2,550,941)
Excluded:
Prevalent users
(N = 2,060,906)
FIGURE 1 Flow-chart of cohort selection.
Franchi et al.
2110 www.jhypertension.com Volume 42 Number 12 December 2024
DISCUSSION
The current study on a large population-based cohort of
patients followed for more than 10 years provides several
findings. The first finding is that ACEIs, ARBs, and thiazide
diuretics did not or only modestly modify the risk of cancer
occurrence both when data were calculated according to
the intention to treat approach and the drug initially pre-
scribed and when an ‘‘as treated’’ approach that took into
account the treatment modifications during the follow-up
TABLE 1. Baseline characteristics of the cohort members
All users ACEIs ARBs Thiazides Thiazides-like Loop D BBs CCBs
N338 910 193 969 144 176 128 337 8616 39 569 118 205 124 342
Age
Median (Q1– Q3) 59 (50– 68) 60 (51– 69) 59 (50– 68) 59 (50– 68) 60 (51– 70) 68 (58 –76) 59 (50 –67) 60 (51 –69)
Age <50 79 051 (23.3) 41 273 (21.3) 32 768 (22.7) 28 422 (22.2) 1776 (20.6) 4332 (11.0) 28 368 (24.0) 27 475 (22.1)
Age 50– 59 95 627 (28.2) 54 247 (28.0) 42 160 (29.2) 37 036 (28.9) 2349 (27.3) 6722 (17.0) 34 055 (28.8) 33 791 (27.2)
Age 60– 69 88 523 (26.1) 53 084 (27.4) 39 593 (27.5) 35 544 (27.7) 2285 (26.5) 10 360 (26.2) 32 233 (27.3) 33 899 (27.3)
Age 70 75 709 (22.4) 45 365 (23.4) 29 655 (20.6) 27 335 (21.3) 2206 (25.6) 18 155 (46.0) 23 549 (19.9) 29 177 (23.5)
Sex
Male 167 841 (49.5) 103 229 (53.2) 74 963 (52.0) 62 760 (48.9) 3344 (38.8) 20 399 (51.5) 57 545 (48.7) 66 345 (53.4)
Female 171 069 (50.5) 90 740 (46.8) 69 213 (48.0) 65 577 (51.1) 5272 (61.2) 19 170 (48.5) 60 660 (51.3) 57 997 (46.6)
MCS
Low (0– 4) 303 787 (89.6) 174 340 (89.9) 133 122 (92.3) 119 702 (93.3) 7948 (92.3) 26 791 (67.7) 104 644 (88.5) 112 792 (90.7)
Intermediate (5– 14) 32 952 (9.7) 18 646 (9.6) 10 595 (7.4) 8269 (6.4) 640 (7.4) 11 678 (29.5) 12 943 (11.0) 10 939 (8.8)
High (>14) 2171 (0.6) 983 (0.5) 459 (0.3) 366 (0.3) 28 (0.3) 1100 (2.8) 618 (0.5) 611 (0.5)
Comorbidities
Respiratory diseases 8806 (2.6) 4925 (2.5) 2898 (2.0) 2218 (1.7) 165 (1.9) 3297 (8.3) 2915 (2.5) 3132 (2.5)
Hepatic diseases 2295 (0.7) 924 (0.5) 536 (0.4) 422 (0.3) 33 (0.4) 1006 (2.5) 743 (0.6) 596 (0.5)
Kidney diseases 1377 (0.4) 878 (0.5) 522 (0.4) 253 (0.2) 28 (0.3) 680 (1.7) 541 (0.5) 685 (0.6)
AMI 11 481 (3.4) 8624 (4.5) 2878 (2.0) 1839 (1.4) 109 (1.3) 2720 (6.9) 8506 (7.2) 3089 (2.5)
Heart failure 21 815 (6.4) 11 170 (5.8) 5687 (3.9) 4042 (3.2) 307 (3.6) 10 784 (23.3) 8738 (7.4) 5852 (4.7)
Arrhythmia 9985 (2.9) 5274 (2.7) 3406 (2.4) 2253 (1.8) 198 (2.3) 3566 (9.0) 5360 (4.5) 3056 (2.5)
Valvular cardiac 2505 (0.7) 1609 (0.8) 680 (0.5) 451 (0.4) 39 (0.5) 1262 (3.2) 1532 (1.3) 631 (0.5)
Vascular cardiac 2093 (0.6) 1319 (0.7) 714 (0.5) 526 (0.4) 44 (0.5) 728 (1.8) 881 (0.8) 915 (0.7)
Cerebrovascular diseases 7315 (2.2) 4911 (2.5) 2542 (1.7) 2211 (1.7) 1323 (1.5) 1783 (4.5) 2303 (2.0) 3270 (2.6)
Co-treatments
Statins 52 946 (15.6) 33 152 (17.1) 21 720 (15.1) 16 661 (13.0) 1047 (12.2) 8650 (21.9) 22 739 (19.2) 18 769 (15.1)
NSAIDS 110 353 (32.6) 63 271 (32.6) 47 391 (32.9) 42 752 (33.3) 3237 (37.6) 16 054 (40.6) 38 154 (32.3) 40 852 (32.9)
Antidiabetic 23 517 (6.9) 15 284 (7.9) 11 178 (7.8) 7771 (6.1) 485 (5.6) 5155 (13.0) 7853 (6.6) 9235 (7.4)
Antiarrhythmic 6342 (1.9) 3203 (1.7) 2324 (1.6) 1550 (1.2) 130 (1.5) 1969 (5.0) 3224 (2.7) 1889 (1.5)
ACEIs, angiotensin converting enzyme inhibitors; AMI, acute myocardial infarction; ARBs, angiotensin receptor blockers; BBs, beta-blockers; CCBs, calcium channel blockers; Loop D,
loop diuretics; MCS, Multisource Comorbidity Score; N, number of patients; NSAIDS, nonsteroidal anti-inflammatory drugs; Q1, first quartile; Q3, third quartile.
FIGURE 2 Association between different antihypertensive drug classes and cancer risk. Data are shown as hazard ratios (HRs) and 95% confidence intervals (CIs) according
to the intention to treat approach, i.e. the drug initially prescribed. Average follow-up was 10.2 years. HRs were adjusted for baseline covariates (see methods and
quantification of risk of mortality by Multiple Comorbidity Score (MCS). Abbreviations for the drug classes as in Table 1.
Antihypertensive drugs and cancer risk
Journal of Hypertension www.jhypertension.com 2111
was employed. This is in line with the results of a recent
individual participant large meta-analysis of randomized
antihypertensive treatment trials (n>260 000 participants)
[17] in which, however, the median follow-up was only
4.2 years and true exposure of the patients to the drugs of
interest was presumably even shorter because adherence to
treatment was not measured and available evidence shows
that it is low in clinical practice [25] and suboptimal in trials
[15]. In our study the duration of the follow-up was longer
than 10 years and the exposure to diuretics and blockers of
the renin angiotensin system during this period was guaran-
teed by measurements of adherence to treatment. This
allows to extend previous conclusions on the absence of
any cancer risk of these drugs to an observation period that
is much more in line with the time needed for a carcino-
genic effect to become clinically manifest.
The second finding of our study is also in line with the
results of the previously mentioned trial-based meta-analy-
sis [17] that use of CCBs is associated with only a modest
increase in cancer risk and that there is no substantial BB-
cancer risk association. In our study, the modest increase in
cancer risk with CCBs was documented either by the
intention-to-treat analysis and by the analysis that more
precisely quantified patients’drug exposure by taking into
FIGURE 3 Association between antihypertensive drug classes and cancer risk. Data are shown as HRs and 95% confidence intervals based on the ‘‘as treated approach’’, i.
e. by taking into account the increasing exposure to drug classes up to >8 years based on availability of prescriptions. HRs were adjust ed for baseline covariates and MCS,
and mutually adjusted for each drug class. Abbreviations as in preceding figures and Table 1.
Franchi et al.
2112 www.jhypertension.com Volume 42 Number 12 December 2024
account treatment changes and adherence over time, with
both approaches showing a small progressive increase in
cancer risk as duration of the exposure increased from <1
to >8 years. The substantial lack of BB-cancer association
was documented by the absence of any cancer risk modifi-
cation with either type of risk calculation, except for an
increased risk after a long duration (>8 years) of BB expo-
sure. This single case of increased risk is likely to reflect a
chance finding, although an increased cancer risk associat-
ed with a particularly long exposure to BBs cannot
be excluded.
The third finding relates to the observation that thiazide-
like and loop diuretics showed a trend for the risk to
increase as the duration of exposure to these drugs in-
creased. However, with both drugs the confidence intervals
were wide (due to a lower number of available patients and
events compared to other drug classes) and the changes
seen at single time points usually showed no statistical
significance. Furthermore, thiazide-like diuretics did not
show an increase in the risk of cancer occurrence in the
intention-to-treat analysis based on first-step drug assump-
tion, which was associated with a small and barely signifi-
cant increase in cancer risk by loop diuretics. Thus, the
overall results of the present study do not consistently
support a cancerogenic influence of diuretics other than
thiazides with their prolonged use, although other long-
term observational studies might be desirable. This negative
conclusion is in line with the few available observational
studies or individual trials, which have usually not reported
an increased cancer risk with thiazide-like or loop diuretics
[15]. Furthermore, no increase in the risk of cancer has been
found in randomized trials on the use of these drugs.
In our study, we did not address the relationship be-
tween antihypertensive drugs and different cancer types or
sites because, although the overall number of cancers was
very large in our cohort, the study was underpowered for
assessing the association between different antihyperten-
sive drug classes and type or site-specific cancers. Indeed,
by considering a two-tailed first type error of 0.05, and a
statistical power of 0.90, the number of cancers needed to
detect a minimum HR of 1.5 for the association between
duration of exposure to a specific drug class and site-
specific cancer risk was 326. This number was greater than
the number of site-specific cancers observed among short-
term (less than six months) or long-term (more than six
months) users of most classes of antihypertensive drugs.
Our study has several elements of strength. First, our
analysis was based on a large and unselected population,
which was made possible because the cost-free or almost
free Italian healthcare system includes virtually all citizens.
Second, the information provided by the drug prescription
database was accurate because pharmacists are required to
report drug dispensations in detail to obtain reimbursement,
and incorrect reports have legal consequences[36]. Third, by
using an ‘‘as-treated’’ approach, we were able to assess the
impact of the entire treatment pathway of all cohort mem-
bers, thus accounting for treatment discontinuation and
changes in therapy. Fourth, the duration of follow-up was
considerably longer than that available in previous studies.
This was particularly the case for the data derived from a
meta-analysis of randomized trials on antihypertensive drugs
in which the reported patient’s follow-up was about 4 years,
but the true exposure was probably less because of the
notoriously limited adherence to treatment. This can hardly
be considered a suitable time for the influence of a carcino-
genic agent to become clinically manifest. Finally, and most
importantly, the results were remarkably convergent be-
tween different types of data analysis, i.e. the intention to
treat analysis based on the initially assumed drug, the ‘‘as
treated’’ analysis that took into account drug changes and
adherence to treatmentduring the follow-up andthe analysis
which removed the possible association between treatment
changes and outcomes.
This study has also limitations. First, the use of antihyper-
tensive drugs was derived from drug prescriptions, which is a
widely used method to estimate adherence to treatment in
large populations [37]. However, this assumes that the num-
ber of days covered by a prescription corresponds with the
number of days of drug use, which is obviously not always
the case [38]. Second,treatments delivered by private services
are not tracked in the drug prescription databases, which
means that in some patients, drug exposure may have been
longer than what was calculated by our analysis. However,
private purchase of drugs provided free of charge by the RHS
is rare and, in particular, nonreimbursable prescriptions of
antihypertensive drugs not included in the databases repre-
sent a very small fraction of the overall antihypertensive drug
prescriptions of the Region [29]. Third, the observational
nature of our study, cannot ‘‘per se’’ exclude confounding,
and the possibility shouldbe considered that the results were
driven by factors which were not measured and thus not
adjusted for in our analysis. Fourth, although use in hyper-
tension is by far predominant, the drugs analyzed are pre-
scribed also for otherdiseases and conditions, which extends
the conclusion beyond the treatment of hypertension.
In conclusion, our large observational study from the
general population of Lombardy reported data on the risk
of cancer occurrence with antihypertensive drug treatment,
which is based on a considerably longer treatment period
than that available from most previous studies, particularly
treatment trials. This study measured drug availability in
single patients throughout the observation period, thus
providing a more precise measure of true drug exposure.
The main results provide no evidence of a sizeable change
in the overall cancer risk for major antihypertensive drug
classes, even when long exposure periods are taken into
account. A modest exposure-related increase in cancer risk
was observed for CCBs and loop diuretics.
ACKNOWLEDGEMENTS
Previous presentations: The results of the current work
were presented in form of abstract at the 32
nd
European
Meeting on Hypertension and Cardiovascular Protection.
Milan (Italy), June 23–26, 2023
Funding: This work was supported by grants from the
Italian Ministry of Education, University, and Research
(‘PRIN’2017, project 2017728JPK) and from the Italian
Ministry of Health (‘Ricerca Finalizzata 2016’, NET-2016-
02363853). The funding sources had no role in the study
design, data collection, data analysis, data interpretation, or
manuscript writing.
Antihypertensive drugs and cancer risk
Journal of Hypertension www.jhypertension.com 2113
Conflicts of interest
G.C. received research support from the European Com-
munity, the Italian Agency of Drugs, and the Italian Ministry
of University and Research. He participated in a variety of
projects funded by pharmaceutical companies (i.e., Novar-
tis, GSK, Roche, AMGEN, and BMS). He also received
honoraria as a member of Roche’s advisory board. All other
authors declare no competing interests.
G.M. has received honoraria from Astra Zeneca, Boeh-
ringer Ingelheim, Daiichi Sankyo, Medtronic, Menarini
Group, Merck, Novartis, Recordati, Sandoz, Sanofi, and
Servier. S.E.K. has received lecture honoraria from Emcure,
Getz, J.B. Pharma, Merck Healthcare KGaA, Sanofi-Aventis
and Vector-Intas.
REFERENCES
1. Pahor M, Guralnik JM, Salive ME, Corti MC, Carbonin P, Havlik RJ. Do
calcium channel blockers increase the risk of cancer? Am J Hypertens
1996; 9:695– 699.
2. Lin SY, Lin CL, Lin CC, Hsu WH, Lin CD, Wang IK, et al. Association
between angiotensin-converting enzyme inhibitors and lung cancer –
a nationwide, population-based, propensity score-matched cohort
study. Cancers (Basel) 2020; 12:E747.
3. Largent JA, McEligot AJ, Ziogas A, Reid C, Hess J, Leighton N, et al.
Hypertension, diuretics and breast cancer risk. J Hum Hypertens 2006;
20:727– 732.
4. Xie Y, Xu P, Wang M, Zheng Y, Tian T, Yang S, et al. Antihypertensive
medications are associated with the risk of kidney and bladder cancer:
a systematic review and meta-analysis. Aging (Albany NY) 2020;
12:1545– 1562.
5. Qi J, An R, Bhatti P, Spinelli JJ, Murphy RA. Antihypertensive medi-
cations and risk of colorectal cancer: a systematic review and meta-
analysis. Cancer Causes Control 2022; 33:801– 812.
6. Shin D, Lee ES, Kim J, Guerra L, Naik D, Prida X. Association between
the use of thiazide diuretics and the risk of skin cancers: a meta-analysis
of observational studies. J Clin Med Res 2019; 11:247– 255.
7. Cao L, Zhang S, Jia CM, He W, Wu LT, Li YQ, et al. Antihypertensive
drugs use and the risk of prostate cancer: a meta-analysis of 21
observational studies. BMC Urol 2018; 18:17.
8. Tang H, Fu S, Zhai S, Song Y, Han J. Use of antihypertensive drugs and
risk of malignant melanoma: a meta-analysis of observational studies.
Drug Saf 2018; 41:161– 169.
9. Li W, Shi Q, Wang W, Liu J, Li Q, Hou F. Calcium channel blockers and
risk of breast cancer: a meta-analysis of 17 observational studies. PLoS
One 2014; 9:e105801.
10. Ioannidis JP, Zhou Y, Chang CQ, Schully SD, Khoury MJ, Freedman
AN. Potential increased risk of cancer from commonly used medica-
tions: an umbrella review of meta-analyses. Ann Oncol 2014; 25:16 –23.
11. Monami M, Filippi L, Ungar A, Sgrilli F, Antenore A, Dicembrini I, et al.
Further data on beta-blockers and cancer risk: observational study and
meta-analysis of randomized clinical trials. Curr Med Res Opin 2013;
29:369– 378.
12. Corrao G, Scotti L, Bagnardi V, Sega R. Hypertension, antihypertensive
therapy and renal-cell cancer: a meta-analysis. Curr Drug Saf 2007;
2:125– 133.
13. Grossman E, Messerli FH, Goldbourt U. Antihypertensive therapy and
the risk of malignancies. Eur Heart J 2001; 22:1343– 1352.
14. Pedersen SA, Gaist D, Schmidt SAJ, H€
olmich LR, Friis S, Pottega
˚rd A.
Hydrochlorothiazide use and risk of nonmelanoma skin cancer: a
nationwide case-control study from Denmark. J Am Acad Dermatol
2018; 78:673– 681.
15. Shao SC, Lai CC, Chen YH, Lai EC, Hung MJ, Chi CC. Associations of
thiazide use with skin cancers: a systematic review and meta-analysis.
BMC Med 2022; 20:228.
16. Bigagli E, Mugelli A, Mancia G. A reverse translational pharmacological
approach to understand the underlying mechanisms of the reported
association between hydrochlorothiazide and nonmelanoma skin can-
cer. J Hypertens 2022; 40:1647– 1649.
17. Copland E, Canoy D, Nazarzadeh M, Bidel Z, Ramakrishnan R, Wood-
ward M, et al. Blood Pressure Lowering Treatment Trialists’Collabo-
ration. Antihypertensive treatment and risk of cancer: an individual
participant data meta-analysis. Lancet Oncol 2021; 22:558– 570.
18. Bangalore S, Kumar S, Kjeldsen SE, Makani H, Grossman E, Wetterslev
J, et al. Antihypertensive drugs and risk of cancer: network meta-
analyses and trial sequential analyses of 324 168 participants from
randomised trials. Lancet Oncol 2011; 12:65– 82.
19. Sipahi I, Debanne SM, Rowland DY, Simon DI, Fang JC. Angiotensin-
receptor blockade and risk of cancer: meta-analysis of randomised
controlled trials. Lancet Oncol 2010; 11:627– 636.
20. Coleman CI, Baker WL, Kluger J, White CM. Antihypertensive medica-
tion and their impact on cancer incidence: a mixed treatment compari-
son meta-analysis of randomized controlled trials. J Hypertens 2008;
26:622– 629.
21. Sipahi I, Chou J, Mishra P, Debanne SM, Simon DI, Fang JC. Meta-
analysis of randomized controlled trials on effect of angiotensin-con-
verting enzyme inhibitors on cancer risk. Am J Cardiol 2011; 108:294–
301.
22. ARB Trialists Collaboration. Effects of telmisartan, irbesartan, valsartan,
candesartan, and losartan on cancers in 15 trials enrolling 138 769
individuals. J Hypertens. 2011;29:623–35.
23. Centers for DiseaseControl and Prevention. MinimumLatency & Types or
Categories of Cancer. Available at: https://www.cdc.gov/wtc/pdfs/poli-
cies/wtchpminlatcancer2014-11-07-508.pdf [Accessed on 13 July 2023].
24. Burnier M, Egan BM. Adherence in hypertension. Circ Res 2019;
124:1124– 1140.
25. Mancia G, Rea F, Corrao G, Grassi G. Two-drug combinations as first-
step antihypertensive treatment. Circ Res 2019; 124:1113– 1123.
26. Andrade SE, Kahler KH, Frech F, Chan KA. Methods for evaluation of
medication adherence and persistence using automated databases.
Pharmacoepidemiol Drug Saf 2006; 15:565– 574; discussion 575– 7.
27. Mancia G, Kreutz R, Brunstr€
om M, Burnier M, Grassi G, Januszewicz A,
Muiesan ML, et al. 2023 ESH Guidelines for the management of arterial
hypertension the Task Force for the management of arterial hyperten-
sion of the European Society of Hypertension: Endorsed by the
International Society of Hypertension (ISH) and the European Renal
Association (ERA). J Hypertens 2023; 41:1874– 2071.
28. Mancia G, Rea F, Ludergnani M, Apolone G, Corrao G. Renin-angio-
tensin-aldosterone system blockers and the risk of COVID-19. N Engl J
Med 2020; 382:2431– 2440.
29. Italian Medicines Agency. National Reports on Medicines use in Italy.
Year 2019. Available at: https://www.aifa.gov.it/documents/20142/
241052/OsMed_2019_Eng.pdf [Accessed 13 July 2023].
30. Corrao G, Rea F, Di Martino M, De Palma R, Scondotto S, Fusco D, et al.
Developing and validating a novel multisource comorbidity score from
administrative data: a large population-based cohort study from Italy.
BMJ Open 2017; 7:e019503.
31. Robins JM, Finkelstein DM. Correcting for noncompliance and depen-
dent censoring in an AIDS clinical trial with inverse probability of
censoring weighted (IPCW) log-rank tests. Biometrics 2000; 56:779–
788.
32. Baek YH, Noh Y, Oh IS, Jeong HE, Filion KB, Lee H, Shin JY. Analytical
approaches to reduce selection bias in as-treated analyses with missing
in-hospital drug information. Drug Saf 2022; 45:1057– 1067.
33. Hern
an MA, Lanoy E, Costagliola D, Robins JM. Comparison of dy-
namic treatment regimes via inverse probability weighting. Basic Clin
Pharmacol Toxicol 2006; 98:237– 242.
34. Xiao Y, Moodie EM, Abrahamowicz M. Comparison of approaches to
weight truncation for marginal structural Cox models. Epidemiol Meth-
ods 2013; 2:1– 20.
35. Westreich D, Greenland S. The Table 2 fallacy: presenting and inter-
preting confounder and modifier coefficients. Am J Epidemiol 2013;
177:292– 298.
36. Strom BL. Overview of automated databases in pharmacoepidemiol-
ogy. John Wiley & Sons, Ltd; 2012.
37. Halpern MT, Khan ZM, Schmier JK, Burnier M, Caro JJ, Cramer J, et al.
Recommendations for evaluating compliance and persistence with
hypertension therapy using retrospective data. Hypertension 2006;
47:1039– 1048.
38. Corrao G, Mancia G. Research strategies in treatment of hypertension:
value of retrospective real-life data. Eur Heart J 2022; 43:3312– 3322.
Franchi et al.
2114 www.jhypertension.com Volume 42 Number 12 December 2024
