Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Jul 1.
Published in final edited form as: Alzheimers Dement. 2020 May 20;16(7):953–964. doi: 10.1002/alz.12090

Cancer and risk of Alzheimer’s disease: Small association in a nationwide cohort study

Anne G Ording a, Erzsébet Horváth-Puhó a, Katalin Veres a, M Maria Glymour b, Mikael Rørth a, Henrik T Sørensen a, Victor W Henderson a,c,d
PMCID: PMC7351601  NIHMSID: NIHMS1577672  PMID: 32432415

Abstract

INTRODUCTION:

Small observational studies with short-term follow-up suggest cancer patients are at reduced risk of Alzheimer’s disease (AD) compared to the general population.

METHODS:

Nationwide cohort study using Danish population-based health registries (1980–2013) with cancer patients (n=949,309) to identify incident diagnoses of AD. We computed absolute reductions in risk attributed to cancer and standardized incidence rate ratios (SIRs) accounting for survival time, comparing observed to expected number of AD cases.

RESULTS:

During up to 34 years of follow-up of cancer survivors, the attributable risk reduction was 1.3 per 10,000 person-years, SIR=0.94 (95% confidence interval 0.92, 0.96). SIRs were similar after stratification by sex, age and cancer stage, and approached the general population for those surviving >10 years.

DISCUSSION:

Inverse associations between cancer and AD were small and diminished over time. Incidence rates in cancer survivors approached those of the general population, suggesting limited association between cancer and AD risk.

Keywords: Neoplasms, Alzheimer Disease, Dementia, Epidemiology, Risk

BACKGROUND

Several observational studies have shown reductions in risk for Alzheimer’s disease (AD) or dementia subsequent to a cancer diagnosis [1–9]. An intriguing explanation involves opposing pathological processes, from uncontrolled cell proliferation in malignancy to neuronal cell death in AD and dementia [10–13]. An alternative explanation for the inverse association is failure to consider the competing risk of death among cancer patients, whose mortality rates are higher compared with individuals without cancer [14]. Moreover, clinicians may be less likely to pursue a dementia diagnosis in patients with life-threatening illness and shortened life expectancy, leading to an inverse detection bias.

A recent cohort study conducted in the state Washington found mixed evidence for an association between cancer and subsequent AD [hazard ratio (HR)=0.95 (95% confidence interval (CI): 0.77–1.17] for prevalent cancer at study inclusion and 0.73 (95% CI: 0.55–0.96) for incident cancer diagnosed during follow-up, with similar results before and after adjustment for a variety of potentially confounding factors [15]. The investigators stratified their analyses by prevalent cancer at study entry, by incident cancer diagnosed during follow-up, and by cancer stage; power was low due to a relatively small sample (n=4,357 participants aged 65+ years) [15]. Other studies have reported similar findings or even greater risk reductions but have also had parallel limitations, lacking stratification by dementia subtypes or cancer stage or including no information on relevant comorbidities [1–7,9,15].

In this nationwide cohort study with 34 years of follow-up, we examined the risk for AD, vascular dementia (VaD), and all-cause dementia in patients with a first primary cancer diagnosis and within specific primary cancer sites and stages compared to risk within the entire Danish general population.

METHODS

Setting and data sources

The Danish health care system provides tax-supported health care to all residents. Nationwide registries track diagnoses, procedures, and vital status for the entire population. These registries can be linked accurately using the unique civil personal registration number assigned to all Danish residents at birth or upon immigration. The Civil Registration System (CRS) records emigration and vital status [16].

The Danish Cancer Registry (DCR) maintains information on all incident cancers diagnosed in Denmark since 1943, including morphology, histology, and stage at diagnosis [17]. The Danish National Patient Registry (DNPR) has recorded all inpatient discharge diagnoses given to patients since 1977 and diagnoses made at hospital outpatient clinics since 1995 [18]. The Psychiatric Central Research Registry (PCRR) has similarly recorded inpatient psychiatric diagnoses given to patients since 1969 and diagnoses made at psychiatric outpatient visits since 1995 [19]. These registries coded diagnoses according to the Eighth Revision of the International Classification of Diseases (ICD-8) until 1993 and according to the ICD-10 starting in 1994 [18]. The diagnostic codes used in this study are provided in Appendix Table A.

Appendix Table A.

Diagnostic codes and categorizations used in the study.

Dementia, overall
  Alzheimer’s disease ICD-8: 290.10, 290.09, ICD-10: F00, G30
  Vascular dementia ICD-8: 293.09, 293.19, ICD-10: F01
  Other dementias ICD-8: 094.19 and 292.09,290.11, 290.18, 290.19, ICD-10: F02, F03, F1x.73 (F10.73 through F19.73), G23.1; G31.0, G31.0A, G31.0B, G31.1, G31.8B, G31.8E, G31.85
Charlson Comorbidity Index
Score 1
 Myocardial infarction ICD–8: 410; ICD–10: I21, I22, I23
 Congestive heart failure ICD–8: 427.09, 427.10, 427.11, 427.19, 428.99, 782.49; ICD–10: I50, I11.0, I13.0, I13.2
 Peripheral vascular disease ICD–8: 440, 441, 442, 443, 444, 445; ICD–10: I70, I71, I72, I73, I74, I77
 Cerebrovascular disease ICD–8: 430–438; ICD–10: I60–I69, G45, G46
 Dementia – excluded ICD–8: 290.09–290.19, 293.09; ICD–10: F00–F03, F05.1, G30
 Chronic pulmonary disease ICD–8: 490–493, 515–518, ICD–10: J40–J47, J60–J67, J68.4, J70.1, J70.3, J84.1, J92.0, J96.1, J98.2, J98.3
 Connective tissue disease ICD–8: 712, 716, 734, 446, 135.99; ICD–10: M05, M06, M08, M09, M30, M31, M32, M33, M34, M35, M36, D86
 Ulcer disease ICD–8: 530.91, 530.98, 531–534; ICD–10: K22.1, K25–K28
 Mild liver disease ICD–8: 571, 573.01, 573.04; ICD–10: B18, K70.0–K70.3, K70.9, K71, K73, K74, K76.0
 Diabetes type 1 and 2 ICD–8: 249.00, 249.06, 249.07, 249.09, 250.00, 250.06, 250.07, 250.09; ICD–10: E10.0, E10.1, E10.9, E11.0, E11.1, E11.9
Score 2
 Hemiplegia ICD–8: 344; ICD–10: G81, G82
 Moderate to severe renal disease ICD–8: 403, 404, 580–583, 584, 590.09, 593.19, 753.10–753.19, 792; ICD–10: 12, I13, N00–N05, N07, N11, N14, N17–N19, Q61
 Diabetes with end organ damage ICD–8: 249.01–249.05, 249.08, 250.01–250.05, 250.08; ICD–10: E10.2–E10.8, E11.2–E11.8
 Any tumor – excluded ICD–8: 140–194; ICD–10: C00–C75
 Leukemia – excluded ICD–8: 204–207; ICD–10: C91–C95
 Lymphoma – excluded ICD–8: 200–203, 275.59; ICD–10: C81–C85, C88, C90, C96
Score 3
 Moderate to severe liver disease ICD–8: 070.00, 070.02, 070.04, 070.06, 070.08, 573.00, 456.00–456.09; ICD–10: B15.0, B16.0, B16.2, B19.0, K70.4, K72, K76.6, I85
Score 6
 Metastatic solid tumor – excluded ICD–8: 195–198, 199; ICD–10: C76–C80
 AIDS ICD–8: 079.83; ICD–10: B21–B24
Cancers
ICD-7 codes for exclusion of prevalent cancer ICD-7: 140–205
  Hodgkin’s lymphoma ICD–10: C81
  Non–Hodgkin’s lymphoma (excl. leukemia, multiple myeloma, and types with cutaneous manifestations) ICD–10: C82–86, C88 (excl. C826, C840, C841, C848, C863, C866, C884B)
  Multiple myeloma and malignant plasma cell neoplasms ICD–10: C90
  Leukemia ICD–10: C91–C95
  Malignant neoplasm of lymphoid, haematopoietic, and related tissues, unspecified ICD–10: C96
  Haematological cancer with cutaneous manifestations C826 (cutaneous follicle center lymphoma)
C840 (Mycosis fungoides)
C841 (Sézary’s disease)
C848 (Cutaneous T–cell lymphoma, unspecified)
C863 (Subcutaneous panniculitis–like T–cell lymphoma)
C866 (Primary cutaneous CD30–positive T–cell proliferations)
C884B (Lymphoma of skin–associated lymphoid tissue (SALT–lymphoma))
  Liver including intrahepatic bile ducts* ICD–10: C22
  Malignant melanoma including those located in anus and anal canal (morphological codes 872–879) ICD–10: C43
  Non–melanoma skin cancers ICD–10: C44
  Kaposi’s sarcoma ICD–10: C46, B210
  Cervix ICD–10: C53
  Anus and anal canal excl. malignant melanomas (morphologic code 872–879) and basal cell cancers (morphologic code 809) ICD–10: C21
  External female genitalia excluding basal cell carcinomas (morphological code 809) ICD–10: C51
  Lip ICD–10: C00
  Tongue ICD–10: C01–02
  Mouth ICD–10: C03–06
  Tonsil and pharynx ICD–10: C09–C13
  Other and poorly specified location in lip, oral cavity, and pharynx ICD–10: C14
  Larynx ICD–10: C32
  Other and poorly specified location in airways and respiratory organs ICD–10: C39
  Oesophagus ICD–10: C15
  Stomach ICD–10: C16
  Colon incl. rectosigmoid junction ICD–10: C18–C19
  Rectum ICD–10: C20
  Pancreas ICD–10: C25
  Lung, bronchus and trachea ICD–10: C33–C34
  Kidney ICD–10: C64
  Renal pelvis ICD–10: C65
  Ureter ICD–10: C66
  Urinary bladder ICD–10: C67
  Salivary gland ICD–10: C07–C08
  Small intestine ICD–10: C17
  Gallbladder and bile ducts ICD–10: C23–C24
  Other and ill-defined cancers of digestive organs ICD–10: C26
  Nasal cavity, middle ear and accessory sinuses ICD–10: C30–C31
  Thymus ICD–10: C37
  Heart and mediastinum ICD–10: C380–383, C388
  Pleura incl. mesothelioma pleura ICD–10: C384, C450
  Bone and articular cartilage ICD–10: C40–C41
  Mesothelioma ICD–10: C45.1–C45.9
  Peripheral nerves and autonomic nervous system ICD–10: C47
  Retroperitoneum and peritoneum, and malignant neoplasm of other connective and soft tissue ICD–10: C48–C49
  Breast ICD–10: C50
  Vagina excluding basal cell carcinomas (morphological code 809) ICD–10: C52
  Uterus ICD–10: C54–C55
  Ovary and fallopian tube ICD–10: C56, C570–574
  Placenta ICD–10: C58
  Other and unspecified female genital organs ICD–10: C577–579
  Penis excluding basal cell carcinomas (morphological code 809) ICD–10: C60
  Prostate ICD–10: C61
  Testis ICD–10: C62
  Other and unspecified cancers in male genital organs excluding basal cell carcinomas (morphological code 809) ICD–10: C63
  Other and unspecified urinary organs ICD–10: C68
  Eye and adnexa ICD–10: C69
  Meninges ICD–10: C70
  Brain including hypophysis, corpus pineale, and ductus craniopharyngealis ICD–10: C71
  Benign neoplasms of brain and meninges ICD-10: DD32, DD33, DD352, DD353, DD354
  Spinal cord, cranial nerves and other parts of central nervous system ICD–10: C72
  Endocrine glands and related structure ICD–10: C73–C75
  Metastasis and unspecified cancer in lymph nodes (only if there is no primary tumor coded) ICD–10: C77–79 (only if there is no primary tumor coded)
  Malignant neoplasm of other, ill defined, or unspecified sites ICD–10: C76, C80
Open in a new tab

Design and study population

We established a cohort of patients with a first incident primary cancer diagnosed during 1980−−2013 recorded in the DCR. While benign neoplasms were not regarded as cancer, benign neoplasms of the brain and meninges were included in the analysis, because these conditions may cause neurologic symptoms similar to those caused by malignancy. We excluded patients with prevalent dementia diagnosed in an inpatient or outpatient setting up to and including the date of cancer diagnosis. We first analyzed all cancer sites in aggregate. We then selected some of the most common cancer sites and those that have been associated with dementia in previous studies [2,4,7,9,15,20]-- bladder, brain, breast, colon, kidney, leukemia, lung, melanoma, non-melanoma skin cancer, pancreatic, and prostate cancer--for separate analysis because of varying biology and prognosis. The large size of the Danish data is a notable strength of this study because it permits analysis by cancer type. Specificity of associations with only some types of cancer might give insight into the biological phenomena underlying the association or implicate particular types of bias, but prior studies have generally been too small to provide statistically precise estimate for particular cancer types. Information on cancer stage at diagnosis was collected for solid tumors. In total 15,270 cancer patients were excluded from the study due to dementia or a diagnosis likely to represent prodromal dementia (mild cognitive impairment or amnestic syndrome) before or on the cancer diagnosis date, leaving 949,309 cancer patients for analysis.

For patients with the selected cancer types (n=679,122), we also included a general population comparison cohort without any prevalent cancer diagnosis and dementia, mild cognitive impairment, or amnestic syndrome and matched them up to 5:1 within one year of birth, sex, and exact index year of cancer diagnosis.

We obtained information on inpatient and outpatient hospital diagnoses of comorbidities included in the Charlson Comorbidity Index (CCI) that were diagnosed before the cancer diagnosis date and information on receipt of chemotherapy within 3 months after cancer diagnosis [21].

Endpoints and follow-up

Patients and their comparators were followed from their cancer diagnosis/index date until a record of incident dementia (either AD, VaD, or all-cause dementia [including unspecified dementia, Pick’s disease, Creutzfeld-Jakob disease, Huntington’s disease, HIV dementia, Parkinson’s disease dementia, Lewy body dementia, progressive supranuclear palsy, and other specified and unspecified diseases]) recorded in the DNPR or the PCRR, death, emigration from Denmark, or 31 December 2013, whichever occurred first.

Analytic variables

Age on the index date was categorized (<50, 50–59, 60–69, 70–79, and ≥80 years) and comorbidity burden was classified using the CCI [21], excluding malignancy and dementia. An individual’s score weights were defined as total scores of 0, 1, 2–3, and ≥4, and conditions comprising the CCI were also examined individually. Cancer was analyzed as all types combined and by the predefined specific cancer sites. Stage at diagnosis for solid tumors was categorized as localized, regional, distant and missing/unknown. Dementia was classified as AD, VaD, and all-cause dementia.

Statistical analysis

We characterized cancer patients according to baseline characteristics. The potential reduction in dementia cases per 10,000 person-years was computed as the expected minus the observed number of cancer cases divided by the person time in days. We calculated standardized incidence rate ratios (SIRs) as a measure of relative risk by comparing observed dementia incidence among cancer patients with that expected, based on single years of age-, sex-, and single calendar-year-standardized incidence in the general population, and stratified by baseline characteristics. Associated 95% CIs were derived using Byar’s approximation, assuming that the observed number of cases in a specific category followed a Poisson distribution. We used exact 95% CIs when the observed number of cancers was less than ten. Analyses were then stratified for cancer survivors by 0–1 year, >1–5 years, >5–10 years, >10—20 years, and >20–34 years of follow-up. To examine potential effects of chemotherapy or radiotherapy, we repeated the analysis for cancer patients diagnosed in 2004 or later receiving any type of chemotherapy by follow-up periods.

A Cox proportional hazard regression analysis was conducted using patients with the pre-specified cancer types separately and compared with their matched general population comparators in a stratified Cox analysis within the matched factors and follow-up periods as described above.

We conducted a number of sensitivity analyses. Our initial coding classified senile dementia (ICD-8) and unspecified dementia (ICD-10) as all-cause dementia. To account for potential miscoding, we conducted sensitivity analyses that re-categorized these as AD rather than all-cause dementia, with no effect on our results. We also conducted an analysis that only included inpatient diagnoses of dementia. We then repeated the analysis including mild cognitive dementia or amnestic syndrome in the dementia outcome. Finally, for the overall estimates we calculated E-values for dementia SIRs using the formula for HRs with rare outcomes (< 15%) [22]. The E-value describes the strength that an unmeasured confounder or set of confounders would need to have with both the exposure and outcome (assuming these two associations were equal) to fully explain the observed-exposure-outcome association [22].

RESULTS

Characteristics of the cancer cohort

A total of 949,309 cancer patients were included in the study (48.3% male), involving 5,242,643 person-years of observation. Median age at cancer diagnosis was 67 years (interquartile range (IQR): 57–76 years). The median follow-up was 3.1 (IQR: 0.70–8.1) years for cancer patients and 7.2 (IQR: 3.3–13.2) years for population comparisons. The majority of cancer patients had no comorbidities at diagnosis (74%) (Table 1).

Table 1.

Characteristics of 949,309 Danish cancer patients diagnosed between 1980 and 2013, and matched cohort of cancer patients and cancer free Danish residents.

All cancer patients Matched cohorts
Patients with specific cancer types* Matched population comparisons
N % N % N %
Total 949,309 100 679,122 100 3,395,597 100
Gender
Female 490,351 52 344,842 51 1,724,208 51
Male 458,958 48 334,280 49 1,671,389 49
Median age (interquartile range) 67 (57, 76) 68 (58, 76) 68 (58, 76)
Age group, years
<50 134,602 14 84,021 12 420,889 12
50–59 157,616 17 111,176 16 555,402 16
60–69 255,269 27 186,794 28 934,335 28
70–79 254,144 27 187,990 28 939,871 28
80+ 147,678 16 109,141 16 545,100 16
Year of diagnosis
1980–1994 346,494 37 232,586 34 1,162,921 34
1995–2001 183,789 19 128,478 19 642,388 19
2002–2009 267,059 28 200,955 30 1,004,773 30
2010–2013 151,967 16 117,103 17 585,515 17
Charlson Comorbidity Index
No comorbidity (CCI= 0) 706,123 74 501,106 74 2,628,460 77
Mild comorbidity (CCI= 1) 154,049 16 112,705 17 500,327 15
Moderate comorbidty (CCI=2–3) 75,150 7.9 55,192 8.1 228,482 6.7
Severe comorbidity (CCI= ≥4) 13,987 1.5 10,119 1.5 38,328 1.1
Specific comorbid condition
Myocardial infarction 39,099 4.1 29,546 4.4 142,097 4.2
Congestive heart failure 31,901 3.4 23,559 3.5 101,655 3.0
Peripheral vascular disease 34,456 3.6 25,828 3.8 100,550 3.0
Cerebrovascular disease 57,248 6.0 42,592 6.3 200,994 5.9
Chronic pulmonary disease 61,189 6.4 46,446 6.8 173,146 5.1
Connective tissue disease 22,618 2.4 16,602 2.4 72,543 2.1
Ulcer disease 38,399 4.0 26,798 3.9 109,417 3.2
Mild liver disease 9797 1.0 6462 1.0 20,312 0.6
Diabetes I and II 38,505 4.1 27,746 4.1 123,291 3.6
Hemiplegia 1867 0.2 1309 0.2 6091 0.2
Moderate to severe renal disease 12,898 1.4 9398 1.4 32,469 1.0
Diabetes with end organ damage 14,642 1.5 10,689 1.6 49,085 1.4
Moderate to severe liver disease 2674 0.3 1795 0.3 4668 0.1
AIDS 647 0.1 232 0.0 709 0.0
Cancer stage†
Localized 298,538 31 298,538 44
Regional 69,692 7.3 69,692 10
Distant 54,675 5.8 54,675 8.1
Missing/Unknown 526,404 56 256,217 38
Receipt of any radiotherapy‡ 51,512 14
Receipt of any chemotherapy‡ 76,079 21
Median age at dementia diagnosis (interquartile range) 83.1 (77.9, 87.5) 83.5 (78.6, 87.7)
Median follow-up time (interquartile range) 2.8 (0.60, 8.0) 3.1 (0.70, 8.1) 7.2 (3.3, 13.2)
Open in a new tab
*

Prostate cancer (n=68,895), Lung cancer (n=103,428), colon cancer (n=67,315), pancreatic cancer (n=22,114), breast cancer (n=109,755), non-melanoma skin cancer (n=195,899), brain cancer (n=12,433), leukemia (n=21,660), melanoma (n=30,586), liver cancer (n=7,824), kidney cancer (n=15,728), bladder cancer (n=23,485).

†

Only solid cancer. Other cancer types are included in the missing/unknown category.

‡

Received within three months of cancer diagnosis. Restricting to patients diagnosed from 2004 onwards due to data availability.

Risk of dementia after cancer

Table 2 shows the observed and expected number of dementia cases with corresponding SIRs and the absolute reduction in risk attributed to cancer per 10,000 person-years by descriptive characteristics and by follow-up period. During 34 years of follow-up, the observed to expected number was 10,048/10,725 cases of AD, 3598/3930 cases of VaD, and 28,544/29,723 cases of all-cause dementia.

Table 2.

Age, sex, and calendar-year standardized incidence rate ratios (SIRs) with 95% confidence intervals (CIs) and absolute reduction in dementia risk attributed to cancer per 10,000 person-years by descriptive characteristic for in 949,309 Danish cancer patients diagnosed between 1980 and 2013.

Alzheimeŕs disease Vascular dementia All-cause dementia
Observed number Expected number SIR (95% CI) Absolute risk reduction Observed number Expected number SIR (95% CI) Absolute risk reduction Observed number Expected number SIR (95% CI) Absolute risk reduction
All 10,048 10,725 0.94 (0.92, 0.96) 1.3 3598 3930 0.92 (0.89, 0.95) 0.6 28,544 29,723 0.96 (0.95, 0.97) 2.2
Sex
Female 6121 6601 0.93 (0.90, 0.95) 1.6 1904 2076 0.92 (0.88, 0.96) 0.6 16,823 17,649 0.95 (0.94, 0.97) 2.7
Male 3927 4124 0.95 (0.92, 0.98) 0.9 1694 1853 0.91 (0.87, 0.96) 0.7 11,721 12,073 0.97 (0.95, 0.99) 1.6
CCI score
None (CCI=0) 7951 8424 0.94 (0.92, 0.96) 1.1 2441 3065 0.80 (0.77, 0.83) 1.4 21,513 23,294 0.92 (0.9 to0.94) 4.0
Low (CCI=1) 1454 1595 0.91 (0.87, 0.96) 2.6 713 597 1.19 (1.11, 1.28) −2.1 4668 4457 1.05 (1.02, 1.08) −3.9
Moderate (CCI=2–3) 573 626 0.91 (0.84, 0.99) 2.7 383 238 1.61 (1.46, 1.78) −7.4 2068 1753 1.18 (1.13, 1.23) −16.0
High (CCI=4+) 70 79 0.89 (0.69, 1.12) 3.5 61 30 2.03 (1.55, 2.61) −12.2 295 218 1.35 (1.20, 1.52) −30.2
Age at cancer diagnosis
0–49 117 149 0.78 (0.65, 0.94) 0.2 56 65 0.86 (0.65, 1.12) 0.1 571 527 1.08 (1.00, 1.18) −0.3
50–59 752 725 1.04 (0.96, 1.11) −0.2 253 298 0.85 (0.75, 0.96) 0.4 2162 2100 1.03 (0.99, 1.07) −0.5
60–69 2353 2414 0.97 (0.94, 1.02) 0.4 975 995 0.98 (0.92, 1.04) 0.1 6812 6809 1.00 (0.98, 1.02) 0.0
70–79 4250 4485 0.95 (0.92, 0.98) 2.4 1482 1623 0.91 (0.87, 0.96) 1.4 11,433 11,904 0.96 (0.94, 0.98) 4.7
80+ 2576 2953 0.87 (0.84, 0.91) 10.2 832 948 0.88 (0.82, 0.94) 3.1 7566 8382 0.90 (0.88, 0.92) 22.0
Year of cancer diagnosis
1980 – 1994 5059 5076 1.00 (0.97, 1.02) 0.1 1516 1657 0.91 (0.87, 0.96) 0.6 12,786 12,958 0.99 (0.97, 1.00) 0.7
1995 – 2001 2124 2285 0.93 (0.89, 0.97) 1.3 1028 1093 0.94 (0.88, 1.00) 0.5 7480 7759 0.96 (0.94, 0.99) 2.3
2002 – 2009 2447 2832 0.86 (0.83, 0.90) 3.0 940 1033 0.91 (0.85, 0.97) 0.7 7157 7756 0.92 (0.90, 0.94) 4.7
2010 – 2013 418 531 0.79 (0.71, 0.87) 4.6 114 147 0.78 (0.64, 0.93) 1.3 1121 1250 0.90 (0.85, 0.95) 5.3
Year of follow-up
0–1 1294 1348 0.96 (0.91, 1.01) 0.7 405 453 0.89 (0.81, 0.99) 0.6 3446 3346 1.03 (1.00, 1.07) −1.3
>1–5 2964 3517 0.84 (0.81, 0.87) 2.9 1080 1219 0.89 (0.83, 0.94) 0.7 8102 9137 0.89 (0.87, 0.91) 5.5
>5–10 2495 2602 0.96 (0.92, 1.00) 0.8 898 963 0.93 (0.87, 1.00) 0.5 7033 7308 0.96 (0.94, 0.99) 2.1
>10–20 2446 2406 1.02 (0.98, 1.06) −0.4 994 1004 0.99 (0.93, 1.05) 0.1 7648 7572 1.01 (0.99, 1.03) −0.7
>20–34 849 852 1.00 (0.93, 1.07) 0.1 221 291 0.76 (0.66, 0.87) 2.8 2315 2361 0.98 (0.94, 1.02) 1.8
Specific comorbidities in the CCI
Myocardial infarction 358 411 0.87 (0.78, 0.97) 4.4 201 167 1.20 (1.04, 1.38) −2.8 1142 1182 0.97 (0.91, 1.02) 3.3
Congestive heart failure 231 288 0.80 (0.70, 0.91) 8.4 136 107 1.27 (1.07, 1.50) −4.3 832 820 1.01 (0.95, 1.09) −1.8
Peripheral vascular disease 250 281 0.89 (0.78, 1.01) 3.6 140 107 1.30 (1.10, 1.54) −3.8 859 779 1.10 (1.03, 1.18) −9.2
Cerebrovascular disease 560 599 0.93 (0.86, 1.02) 2.4 532 223 2.39 (2.19, 2.60) −18.9 2250 1686 1.33 (1.28, 1.39) −34.5
Chronic pulmonary disease 385 461 0.83 (0.75, 0.92) 4.5 153 171 0.89 (0.76, 1.05) 1.1 1247 1251 1.00 (0.94, 1.05) 0.2
Connective tissue disease 220 250 0.88 (0.77, 1.00) 3.7 103 88 1.16 (0.95, 1.41) −1.8 691 692 1.00 (0.93, 1.08) 0.2
Ulcer disease 384 365 1.05 (0.95, 1.16) −1.6 177 142 1.24 (1.07, 1.44) −2.8 1220 1051 1.16 (1.10, 1.23) −13.7
Mild liver disease 34 35 0.98 (0.68, 1.37) 0.3 20 13 1.49 (0.91, 2.29) −2.7 165 97 1.70 (1.45, 1.98) −28.1
Diabetes I and II 314 310 1.01 (0.91, 1.13) −0.4 192 110 1.74 (1.50, 2.01) −7.4 1085 822 1.32 (1.24, 1.40) −23.8
Hemiplegia 5 13 0.40 (0.13, 0.93) 13.8 <5 4 NA 0.5 29 33 0.88 (0.59, 1.27) 7.0
Moderate to severe renal disease 81 92 0.88 (0.70, 1.09) 2.9 51 35 1.46 (1.09, 1.92) −4.3 285 256 1.11 (0.99, 1.25) −7.6
Diabetes with end organ 101 105 0.96 (0.79, 1.17) 1.0 87 40 2.19 (1.76, 2.71) −12.6 397 287 1.38 (1.25, 1.52) −29.2
Moderate to severe liver disease <5 7 NA NA <5 3 NA NA 26 20 1.29 (0.84, 1.90) −10.7
AIDS <5 1 NA NA <5 0 NA NA 5 2 2.57 (0.83, 5.98) −14.3
Cancer stage*
Localized 733 824 0.89 (0.83, 0.96) 2.8 253 272 0.93 (0.82, 1.05) 0.6 1910 2156 0.89 (0.85, 0.93) 7.5
Regional 6912 7195 0.96 (0.94, 0.98) 0.8 2452 2670 0.92 (0.88, 0.96) 0.6 19,503 20,192 0.97 (0.95, 0.98) 1.9
Distant 585 691 0.85 (0.78, 0.92) 2.7 176 243 0.73 (0.62, 0.84) 1.7 1677 1812 0.93 (0.88, 0.97) 3.4
Missing 1818 2014 0.90 (0.86, 0.94) 2.1 717 745 0.96 (0.89, 1.04) 0.3 5454 5562 0.98 (0.95, 1.01) 1.2
Radiotherapy by follow-up period†
>3 months-10 years 165 220 0.75 (0.64, 0.87) 4.1 55 70 0.79 (0.60, 1.03) 1.1 494 545 0.91 (0.83, 0.99) 3.8
>3months-1 year 23 42 0.55 (0.35, 0.82) 6.5 12 15 0.80 (0.41, 1.39) 1.1 98 111 0.88 (0.72, 1.08) 4.5
>1–5 years 95 129 0.74 (0.60, 0.90) 4.1 38 41 0.92 (0.65, 1.26) 0.4 293 322 0.91 (0.81, 1.02) 3.6
>5–10 years 47 49 0.96 (0.70, 1.27) 0.9 5 13 0.38 (0.12, 0.89) 3.4 103 113 0.91 (0.75, 1.11) 4.0
Chemotherapy by follow-up period†
>3 months-10 years 130 170 0.76 (0.64, 0.91) 2.2 30 55 0.55 (0.37,0.78) 1.4 342 425 0.80 (0.72, 0.89) 4.6
>3months-1 year 18 41 0.44 (0.26, 0.70) 5.2 <10 15 0.20 (0.04,0.60) 2.7 70 107 0.65 (0.51–0.83) 8.4
>1–5 years 75 98 0.77 (0.60, 0.96) 2.1 20 32 0.63 (0.38, 0.97) 1.1 197 245 0.80 (0.70–0.92) 4.5
>5–10 years 37 32 1.16 (0.82, 1.60) −1.8 <10 9 0.82 (0.33, 1.68) 0.6 75 73 1.02 (0.81–1.28) 0.6
Open in a new tab

CCI. Charlson comorbidity index.

*

Only solid tumors

†

Treatment received within three months of cancer diagnosis, starting follow-up at this date. Restricting to patients diagnosed from 2004 onwards due to data availability.

After any cancer diagnosis, we observed a modestly reduced SIR for all-cause dementia (SIR: 0.96, 95% CI: 0.95–0.97). The absolute reduction in dementia risk attributed to cancer was 2.2 per 10,000 persons. Similar reductions were observed for AD and VaD and for males and females. SIRs for all-cause dementia and AD were above 1.0 for individuals with higher CCI scores and for individuals with most individual comorbid conditions, but were under 1.0 (SIR=0.91 [ 95% CI: 0.84–0.99]) for patients with comorbidity scores of 2–3. Stratified by age group at cancer diagnosis or cancer treatment with chemotherapy or radiotherapy, the risks for all-cause dementia, AD, or VaD were near unity (Table 2). When stratified by follow-up period, the long-term (34-year) relative risk of dementia among cancer survivors approached a null association for AD and all-cause dementia, although not for VaD. The inverse associations between cancer and AD and all-cause dementia were somewhat more pronounced for a cancer diagnosis in recent years (from 1995 onwards) than in earlier years (Table 2). In a sensitivity analysis including only inpatient diagnoses, the SIR for all-cause dementia was 0.94 (95% CI: 0.93–0.95), compared to 0.96 (95% CI: 0.95–0.97) in the primary analysis. In the analysis that added mild cognitive impairment and amnestic syndrome to the definition of dementia the number of observed all-cause dementia cases increased to 29,745 from 28,544 in our primary analysis and the association between cancer and dementia was similar to the primary analysis (SIR=0.97; 95% CI: 0.96–0.98).

Based on SIRs for the primary analysis, we estimate E-values as 1.32 for AD, 1.39 for VaD, and 1.25 for all-cause dementia.

We found diverging results by specific cancer sites, though the stage stratified analyses were restricted by low number of dementia cases. The SIRs were elevated for all types of dementia after brain cancer; increased for all-cause dementia but null or reduced for AD and VaD after lung cancer; and reduced or similar to that of the general population for most of the remaining cancer sites (Table 3).

Table 3.

Standardized incidence rate ratios (SIRs) with 95% confidence intervals (CIs) and absolute reduction in dementia risk attributed to cancer per 10,000 person-years for any dementia, in 679,122 Danish cancer patients with specific cancer sites diagnosed between 1980 and 2013, by stage.

Alzheimer’s disease Vascular dementia All-cause dementia
Observed number Expected number SIR (95% CI) Absolute risk reduction Observed number Expected number SIR (95% CI) Absolute risk reduction Observed number Expected number SIR (95% CI) Absolute risk reduction
Bladder cancer 227 242 0.94 (0.82, 1.07) 1.6 98 97 1.01 (0.82, 1.23) −0.1 640 662 0.97 (0.89, 1.04) 2.3
Localized 135 148 0.91 (0.77, 1.08) 2.1 59 60 0.99 (0.75, 1.28) 0.1 385 403 0.96 (0.86, 1.06) 3.0
Regional 28 23 1.23 (0.81, 1.77) −5.2 17 9 1.90 (1.11, 3.05) −8.1 67 60 1.11 (0.86, 1.41) −6.8
Distant <5 3 NA 12.9 <5 1 NA 7.6 8 7 1.08 (0.47, 2.13) −3.9
Missing/Unknown 63 69 0.92 (0.70, 1.17) 2.5 22 28 0.79 (0.50, 1.20) 2.5 180 192 0.94 (0.81, 1.09) 5.0
Brain cancer 13 8 1.62 (0.86, 2.77) −1.6 10 3 2.90 (1.39, 5.33) −2.1 91 24 3.77 (3.03, 4.63) −21.1
Breast cancer 1495 1569 0.95 (0.91, 1.00) 0.8 486 492 0.99 (0.90, 1.08) 0.1 4129 4132 1.00 (0.97, 1.03) 0.0
Localized 899 901 1.00 (0.93, 1.07) 0.0 276 285 0.97 (0.86, 1.09) 0.2 2403 2385 1.01 (0.97, 1.05) −0.4
Regional 419 471 0.89 (0.81, 0.98) 1.7 148 147 1.01 (0.85, 1.18) 0.0 1176 1214 0.97 (0.91, 1.03) 1.2
Distant 30 38 0.79 (0.53, 1.12) 4.3 10 11 0.90 (0.43, 1.66) 0.6 94 95 0.99 (0.80, 1.21) 0.4
Missing/unknown 147 158 0.93 (0.78, 1.09) 1.9 52 50 1.05 (0.78, 1.37) −0.4 456 438 1.04 (0.95, 1.14) −3.0
Colon cancer 833 944 0.88 (0.82, 0.94) 3.6 279 337 0.83 (0.73, 0.93) 1.9 2343 2598 0.90 (0.87, 0.94) 8.4
Localized 207 253 0.82 (0.71, 0.94) 6.0 71 86 0.83 (0.64, 1.04) 1.9 585 668 0.88 (0.81, 0.95) 10.8
Regional 93 104 0.90 (0.73, 1.10) 3.0 22 34 0.64 (0.40, 0.98) 3.4 231 266 0.87 (0.76, 0.99) 9.9
Distant 20 27 0.75 (0.46, 1.16) 5.8 <5 8 NA 4.8 54 64 0.85 (0.64, 1.11) 8.3
Missing/unknown 513 561 0.91 (0.84, 1.00) 2.6 183 208 0.88 (0.76, 1.02) 1.4 1473 1601 0.92 (0.87, 0.97) 7.0
Kidney cancer 86 110 0.78 (0.62, 0.96) 3.7 53 42 1.25 (0.94, 1.63) −1.6 286 303 0.95 (0.84, 1.06) 2.5
Localized 62 74 0.84 (0.65, 1.08) 2.6 43 29 1.48 (1.07, 2.00) −3.1 204 206 0.99 (0.86, 1.14) 0.4
Regional 5 9 0.56 (0.18, 1.30) 8.1 <5 3 NA NA 17 24 0.72 (0.42, 1.15) 13.4
Distant 6 7 0.86 (0.31, 1.86) 2.0 <5 3 NA NA 16 18 0.88 (0.50, 1.43) 4.2
Missing/unknown 13 21 0.63 (0.33, 1.08) 7.6 8 8 1.06 (0.46, 2.08) −0.4 49 55 0.89 (0.66, 1.18) 6.0
Leukemia 150 162 0.93 (0.78, 1.09) 1.2 51 60 0.85 (0.63, 1.11) 0.9 395 440 0.90 (0.81, 0.99) 4.5
Lung cancer 174 207 0.84 (0.72, 0.97) 2.5 65 81 0.80 (0.62, 1.02) 1.2 613 546 1.12 (1.04, 1.22) −5.1
Localized 83 101 0.82 (0.66, 1.02) 2.9 38 41 0.93 (0.66, 1.28) 0.5 291 272 1.07 (0.95, 1.20) −3.0
Regional 32 43 0.75 (0.51, 1.06) 3.2 12 18 0.68 (0.35, 1.19) 1.7 124 115 1.07 (0.89, 1.28) −2.6
Distant 25 38 0.67 (0.43, 0.98) 5.4 8 14 0.58 (0.25, 1.14) 2.5 110 97 1.14 (0.93, 1.37) −5.6
Missing/unknown 34 26 1.31 (0.91, 1.83) −6.5 7 9 0.82 (0.33, 1.68) 1.3 88 61 1.43 (1.15, 1.76) −21.6
Melanoma skin cancer 284 325 0.87 (0.78, 0.98) 1.5 92 124 0.74 (0.60, 0.91) 1.2 805 929 0.87 (0.81, 0.93) 4.5
Localized 253 285 0.89 (0.78, 1.00) 1.3 82 110 0.75 (0.59, 0.93) 1.1 715 820 0.87 (0.81, 0.94) 4.2
Regional 9 10 0.94 (0.43, 1.78) 0.7 <5 4 NA NA 27 27 1.01 (0.66, 1.47) −0.3
Distant <5 2.8 NA NA <5 1 NA NA <5 7 NA NA
Missing/unknown 20 27 0.73 (0.44, 1.12) 3.9 9 10 0.91 (0.42, 1.74) 0.4 59 76 0.78 (0.59, 1.00) 8.9
Non-melanoma skin cancer 3876 4005 0.97 (0.94, 1.00) 0.8 1366 1505 0.91 (0.86, 0.96) 0.9 10805 11479 0.94 (0.92, 0.96) 4.3
Localized 3316 3384 0.98 (0.95, 1.01) 0.5 1171 1277 0.92 (0.87, 0.97) 0.8 9251 9725 0.95 (0.93, 0.97) 3.6
Regional 16 11 1.50 (0.86, 2.43) −15.3 <5 4 NA NA 36 31 1.17 (0.82, 1.63) −15.3
Distant 5 2 2.48 (0.80, 5.77) −35.6 <5 1 NA NA 9 6 1.60 (0.74, 3.05) −40.5
Missing/unknown 539 608 0.89 (0.81, 0.96) 2.8 190 223 0.85 (0.74, 0.98) 1.3 1509 1718 0.88 (0.83, 0.92) 8.5
Pancreatic cancer 19 26 0.72 (0.43, 1.13) 5.2 4 9 0.45 (0.12, 1.15) 3.5 59 66 0.90 (0.69, 1.16) 4.7
Localized 8 7 1.19 (0.51, 2.34) −3.4 <5 2 NA NA 17 17 1.00 (0.58, 1.60) 0.1
Regional <5 6 NA NA <5 2 NA NA 10 14 0.70 (0.34, 1.29) 10.8
Distant <5 5 NA NA <5 2 NA NA 14 11 1.22 (0.67, 2.05) −8.3
Missing/unknown 6 9 0.67 (0.25, 1.46) 9.0 2 3 0.66 (0.08, 2.39) 3.2 18 23 0.79 (0.47, 1.25) 14.9
Prostate cancer 779 813 0.96 (0.89, 1.03) 1.2 325 340 0.96 (0.85, 1.07) 0.5 2175 2237 0.97 (0.93, 1.01) 2.2
Localized 387 399 0.97 (0.88, 1.07) 0.8 162 164 0.99 (0.84, 1.15) 0.1 1080 1077 1.00 (0.94, 1.06) −0.2
Regional 29 27 1.09 (0.73, 1.57) −2.1 10 11 0.91 (0.43, 1.67) 0.9 69 70 0.98 (0.76, 1.24) 1.2
Distant 84 94 0.90 (0.72, 1.11) 3.1 22 40 0.55 (0.35, 0.83) 5.7 217 258 0.84 (0.73, 0.96) 13.0
Missing/unknown 279 294 0.95 (0.84, 1.07) 1.8 131 125 1.05 (0.88, 1.24) −0.7 809 831 0.97 (0.91, 1.04) 2.5
Open in a new tab

Results from the Cox regression analysis of the matched cohorts (i.e., using both the cancer cohort and the general population comparison cohort) and stratified by years of follow-up resembled that of the SIR analysis (data not shown).

DISCUSSION

In this large nationwide cohort study—which involved 34 years (5,242,643 person-years) of follow-up—we observed a 6% lower incidence rate for diagnosis of AD, and 4% for all-cause dementia associated with prior cancer diagnosis. The magnitude of the association was closer to the null than previously reported and further attenuated after 10 years, approaching that of the general population.

During 34 years of follow-up of cancer survivors, the reduction in the number of AD cases attributed to cancer was only 1.3 AD cases per 10,000 person-years. We observed similar associations for VaD, where the primary underlying etiology is less likely to be neurodegenerative, and for dementia of any cause. The magnitude of these associations was considerably less than previously reported, as discussed below.

Strengths and weaknesses

Strengths of this population-based cohort study include its large size and length of follow-up. Furthermore, all Danish hospitals and hospital clinics report data on diagnoses including dementia to the DNPR and the PCRR [18,19]. Data in the DCR are nearly complete and valid due to compulsory reporting except for non-melanoma skin cancer. Most tumors are histologically confirmed since 2009 [17]. Follow-up of all study participants was virtually complete, as patients were identified using comprehensive hospital-based registries in a healthcare system providing free access to health care, and as patients could be tracked easily through the CRS [16,18]. The positive predictive value of the dementia diagnosis in the DNPR overall has been shown to be 89% and is higher for AD than for other dementia types [23]. As many patients with dementia have mixed pathologies, including those of both AD and VaD [24], any misclassification may result in a bias with an unknown direction on the study results for individual causes of dementia, but not for all-cause dementia. To account for potential inaccurate coding, we conducted sensitivity analyses that included the ICD-8 code for senile dementia and the ICD-10 code for dementia unspecified as AD, with no effect on our results. Unfortunately, we did not have information on biomarkers relevant to diagnosis of dementia types.

Several limitations should be considered. As cancer diagnostics and treatment changed over the study period, cancer survivorship may have differed by calendar year of inclusion. However, we stratified our results by calendar year of cancer diagnosis and by follow-up period without changing the results. We excluded cancer patients with a baseline diagnosis of mild cognitive impairment or amnestic syndrome to reduce inclusion of patients with prodromal dementia. Dementia and cancer are both associated with a latency and prodromal period before diagnosis [25]. For AD, the prodromal period may be as long as 25 years [26]. The temporal indeterminacy of disease onset is a concern for shorter-term associations. Although competing risk of death is a concern when comparing cumulative risk of dementia among individuals with a cancer history to cumulative risk of dementia among individuals without a cancer history, it is unlikely to explain our results because our analyses are based on incidence rates, which intrinsically correct for differences in time at risk. Residual confounding is a possible explanation and -- unlike in typical studies focused on causal effects of an exposure on an outcome-- would be an important explanation in this case. A confounder-based explanation of the inverse association between cancer and dementia implies that there is an unmeasured factor that increases cancer risk but decreases dementia risk (or vice-versa). This is an important possible finding, and would justify further research on what that unmeasured factor might be. Diagnostic bias is also a possible explanation for our results. Recent work in a large population-based cohort found that individuals subsequently diagnosed with cancer had slightly better memory functioning even prior to cancer diagnosis [27]. This would imply that the diagnostic bias, if relevant, results from individuals with higher cognitive function having increased chance of cancer diagnosis. We consider this most plausible for cancers that are often undiagnosed or diagnosed upon screening, such as prostate and breast cancer.

Previous literature

Most studies that assessed the risk for AD or dementia after cancer were conducted in the United States, were limited to study populations aged 65+ years, and were not population based [1–7,9,15]. The reported inverse associations between cancer and dementia is stronger for AD than for other dementia types. For example, Roe et al. reported that cancer was associated with a reduced risk for AD (HR= 0.57, 95% CI: 0.36–0.90) after adjustment for a number of factors, including APOE genotype, over a median of 5.4 years of follow-up. No association was found between cancer and VaD [4]. Driver et al. found a reduced risk of probable AD (HR=0.67, 95% CI: 0.47–0.97) following a cancer diagnosis, particularly for survivors of lung cancer and other smoking-related cancer (HR=0.26, 95% CI 0.08–0.82) [2]. Freedman et al. similarly reported a decreased risk for AD within 10 years of follow-up after diagnosis of cancer at different sites [7]. In a study with 21 years follow-up Bowles et al. reported a HR for AD of 0.95 (95% CI: 0.77–1.17) for patients with prevalent cancer and 0.73 (95% CI: 0.55–0.96) for patients with incident cancer [15]. Findings from Frain et al. based on data from the US Veteran Healthcare System including patients age 65+ years, support previous studies of an inverse association between some cancer types and subsequent AD [9]. A study conducted in Italy reported similar findings in a population aged 60+ years [3]. However, a study based on data from the Utah Population Database and argue that such associations arise from bias due to the competing risk of dying [14]. Both for AD and for VaD, we found a lower relative risk associated with some smoking- and alcohol-related cancers such as those of the lung and colon, but not for other smoking or alcohol-related cancers (bladder, breast, kidney, pancreatic cancer), and not for long-term cancer survivors.

Potential Mechanism

Neurodegenerative diseases and carcinogenesis may share several molecular signaling pathways and factors involved in cell cycle dysfunction [8,13,28–33]. Though still controversial, chemotherapeutic agents may affect cognition, ranging from maintaining cognitive function through suppressed inflammation and blocked cell cycles, on the one hand [8], to potential peripheral neurotoxic effects, on the other hand [12]. As well, specific comorbidities may lead to differing diagnostic procedures and treatment choices [8,34,35]. We found inverse associations between cancer and AD across all comorbidity levels, but increased risk for VaD and all-cause dementia among cancer patients with increasing CCI scores.

We also found an initial inverse association between some smoking- and alcohol-related cancers and AD and VaD, which diminished during follow-up for cancer survivors. Patients with disseminated cancer are likely to have died from their disease in the first decade after diagnosis, leaving disease-free cancer patients for the long-term analysis. Misclassification of mild AD or mild dementia as no-dementia in patients with cancer also could explain the small, early inverse association. We found an increased risk of dementia after diagnosis of both benign and malignant brain tumor, consistent with the well-known increase in dementia and cognitive dysfunction associated with all types of brain tumors.

Our overall results for AD and VaD, which are largely similar, contrast with evidence that separated AD from other types of dementia, such as the results of Roe [4] Driver [2] and Freedman [7], although other studies have reported a decreased risk for all-cause dementia after cancer [1,6]. It is possible that neurodegeneration among cancer patients with comorbid cardiovascular conditions tended to be classified as VaD rather than AD, especially since the dementia diagnoses were not supported by brain imaging for the majority of our registry-defined cohort. Therefore, some patients classified as having VaD may have had Alzheimer’s pathology or a combination of vascular and AD pathologies [36]. When we stratified results by cancer stage, we also found largely similar results by stage category within each dementia subtype. This suggests that dementia is not underdiagnosed in patients with metastatic cancer who have poorer survival prospects than the general population.

Conclusions and implications

The evidence presented here augments recent reports showing that cancer patients may have decreased relative and absolute risks for AD after standardization to the general population. However, the magnitude of the association was much less than previously reported and diminished after 10 years, approaching that found in the general population. This study does not indicate a clinically relevant association between cancer and risk of AD, VaD, or all-cause dementia. Interest in the association is largely because it suggests the possibility of a physiologic process or risk factor that increases cancer risk but decreases dementia risk, i.e., a potential confounder with biological relevance; identifying such a potential confounder might provide novel insight into the mechanisms of dementia. When interpreting the strength of the observed inverse association, it is therefore valuable to assess how strong the effects of such a potential confounder or set of confounders would have to be in order to account for the association. The E-value provides such an estimate: under the assumption that the confounder has equally strong (but inverse) associations with cancer and dementia, the confounder or set of confounders would need to increase dementia risk by roughly 25% to account for the observed inverse association. The modest effect estimates are not as large as for single major risk factors such as APOE-e4 allele but are in line with estimates related to some cardiovascular risk factors, for example. Such small effects may be of biological interest if they relate to pathological cascade that culminate in dementia, but could also plausibly be explained by detection bias. We also note that the major advantages of our use of administrative records, which provides a much larger sample size than ever previously evaluated, also introduces potential biases that may underestimate the association between cancer and dementia, for example due to non-differential misdiagnoses.

Contributorship statement

HTS conceived the study idea and designed the study together with AGO and VWH. EHP and KV carried out the analyses. AGO organized the writing and wrote the initial draft. All authors participated in the discussion and interpretation of the results. All authors critically revised the manuscript for intellectual content and approved the final version before submission. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. AGO is the guarantor.

Acknowledgments

Source of funding

The study was supported by a grant from Lundbeckfonden (grant no. R248–2017-521), Aarhus University Hospital and the Department of Clinical Epidemiology’s Research Foundation. MNG was supported by National Institutes of Health grant RF1 AG059872. VWH was supported by National Institutes of Health grant P50 AG047366. The funding sources had no role in the design, conduct, analysis, or reporting of the study.

Abbreviations

AD

Alzheimer’s disease

HR

Hazard ratio

VaD

Vascular dementia

CRS

Civil Registration System

DCR

Danish Cancer Registry

DNPR

Danish National Patient Registry

PCRR

Psychiatric Central Research Registry

ICD

International Classification of Disease

CCI

Charlson comorbidity Index

SIR

Standardized incidence rate ratio

CI

Confidence interval

IQR

Interquartile range

Footnotes

Declarations of interest:

None.

Data permission

The study was approved by the Danish Data Protection Agency (record number 2007–58-0010).

Ethics committee approval: No ethical committee approval was needed.

Data sharing: Not allowed.

Patient involvement: No patient involvement.

References

  • [1].Attner B, Lithman T, Noreen D, Olsson H. Low cancer rates among patients with dementia in a population-based register study in Sweden. Dement Geriatr Cogn Disord 2010;30:39–42. 10.1159/000315509 [doi]. [DOI] [PubMed] [Google Scholar]
  • [2].Driver JA, Beiser A, Au R, Kreger BE, Splansky GL, Kurth T, et al. Inverse association between cancer and Alzheimer’s disease: results from the Framingham Heart Study. BMJ 2012;344:e1442 10.1136/bmj.e1442 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [3].Musicco M, Adorni F, Di Santo S, Prinelli F, Pettenati C, Caltagirone C, et al. Inverse occurrence of cancer and Alzheimer disease: a population-based incidence study. Neurology 2013;81:322–8. 10.1212/WNL.0b013e31829c5ec1 [doi]. [DOI] [PubMed] [Google Scholar]
  • [4].Roe CM, Fitzpatrick AL, Xiong C, Sieh W, Kuller L, Miller JP, et al. Cancer linked to Alzheimer disease but not vascular dementia. Neurology 2010;74:106–12. 10.1212/WNL.0b013e3181c91873 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [5].Ou SM, Lee YJ, Hu YW, Liu CJ, Chen TJ, Fuh JL, et al. Does Alzheimer’s disease protect against cancers? A nationwide population-based study. Neuroepidemiology 2013;40:42–9. 10.1159/000341411 [doi]. [DOI] [PubMed] [Google Scholar]
  • [6].Lin HL, Lin HC, Tseng YF, Chen SC, Hsu CY. Inverse Association Between Cancer and Dementia: A Population-based Registry Study in Taiwan. Alzheimer Dis Assoc Disord 2016;30:118–22. 10.1097/WAD.0000000000000116 [doi]. [DOI] [PubMed] [Google Scholar]
  • [7].Freedman DM, Wu J, Chen H, Kuncl RW, Enewold LR, Engels EA, et al. Associations between cancer and Alzheimer’s disease in a U.S. Medicare population. Cancer Med 2016; 5:2965–2976. 10.1002/cam4.850 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [8].Akushevich I, Kravchenko J, Ukraintseva S, Arbeev K, Kulminski A, Yashin AI. Morbidity risks among older adults with pre-existing age-related diseases. Exp Gerontol 2013;48:1395–401. . [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [9].Frain L, Swanson D, Cho K, Gagnon D, Lu KP, Betensky RA, et al. Association of cancer and Alzheimer’s disease risk in a national cohort of veterans. Alzheimers Dement 2017;13:1364–70. 10.1016/j.jalz.2017.04.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [10].Ibanez K, Boullosa C, Tabares-Seisdedos R, Baudot A, Valencia A. Molecular evidence for the inverse comorbidity between central nervous system disorders and cancers detected by transcriptomic meta-analyses. PLoS Genet 2014;10:e1004173 10.1371/journal.pgen.1004173 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [11].Driver JA, Lu KP. Pin1: a new genetic link between Alzheimer’s disease, cancer and aging. Curr Aging Sci 2010;3:158–65. https://doi.org/BSP/CAS/E-Pub/00009 [pii]. [DOI] [PubMed] [Google Scholar]
  • [12].Ganguli M. Cancer and Dementia: It’s Complicated. Alzheimer Dis Assoc Disord 2015;29:177–82. 10.1097/WAD.0000000000000086 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [13].Tabares-Seisdedos R, Dumont N, Baudot A, Valderas JM, Climent J, Valencia A, et al. No paradox, no progress: inverse cancer comorbidity in people with other complex diseases. Lancet Oncol 2011;12:604–8. 10.1016/S1470-2045(11)70041-9 [doi]. [DOI] [PubMed] [Google Scholar]
  • [14].Hanson HA, Horn KP, Rasmussen KM, Hoffman JM, Smith KR. Is Cancer Protective for Subsequent Alzheimer’s Disease Risk? Evidence From the Utah Population Database. Journals Gerontol Ser B Psychol Sci Soc Sci 2016;72:gbw040 10.1093/geronb/gbw040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [15].Bowles EJA, Walker RL, Anderson ML, Dublin S, Crane PK, Larson EB. Risk of Alzheimer’s disease or dementia following a cancer diagnosis. PLoS One 2017;12:e0179857 10.1371/journal.pone.0179857 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [16].Schmidt M, Pedersen L, Sorensen HT. The Danish Civil Registration System as a tool in epidemiology. Eur J Epidemiol 2014;29:541–9. 10.1007/s10654-014-9930-3 [doi]. [DOI] [PubMed] [Google Scholar]
  • [17].Gjerstorff ML. The Danish Cancer Registry. Scand J Public Health 2011;39:42–5. 10.1177/1403494810393562. [DOI] [PubMed] [Google Scholar]
  • [18].Schmidt M, Schmidt SA, Sandegaard JL, Ehrenstein V, Pedersen L, Sorensen HT. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol 2015;7:449–90. 10.2147/CLEP.S91125 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [19].Mors O, Perto GP, Mortensen PB. The Danish Psychiatric Central Research Register. Scand J Public Health 2011;39:54–7. 10.1177/1403494810395825. [DOI] [PubMed] [Google Scholar]
  • [20].Schmidt SAJ, Ording AG, Horváth-Puhó E, Sørensen HT, Henderson VW. Non-melanoma skin cancer and risk of Alzheimer’s disease and all-cause dementia. PLoS One 2017;12:e0171527 10.1371/journal.pone.0171527. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [21].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. [DOI] [PubMed] [Google Scholar]
  • [22].VanderWeele TJ, Ding P. Sensitivity Analysis in Observational Research: Introducing the E-Value. Ann Intern Med 2017; 167(4):268–274. 10.7326/M16-2607 [doi]. [DOI] [PubMed] [Google Scholar]
  • [23].Phung TK, Andersen BB, Kessing L V, Mortensen PB, Waldemar G. Diagnostic evaluation of dementia in the secondary health care sector. Dement Geriatr Cogn Disord 2009;27:534–42. 10.1159/000223664 [doi]. [DOI] [PubMed] [Google Scholar]
  • [24].Schneider JA, Arvanitakis Z, Bang W, Bennett DA. Mixed brain pathologies account for most dementia cases in community-dwelling older persons. Neurology 2007;69:2197–204. https://doi.org/01.wnl.0000271090.28148.24 [pii]. [DOI] [PubMed] [Google Scholar]
  • [25].Wilson RS, Leurgans SE, Boyle PA, Bennett DA. Cognitive decline in prodromal Alzheimer disease and mild cognitive impairment. Arch Neurol 2011;68:351–6. 10.1001/archneurol.2011.31 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [26].Bateman RJ, Xiong C, Benzinger TL, Fagan AM, Goate A, Fox NC, et al. Clinical and biomarker changes in dominantly inherited Alzheimer’s disease. N Engl J Med 2012;367:795–804. 10.1056/NEJMoa1202753 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [27].Ospina-Romero M, Abdiwahab E, Kobayashi L, Filshtein T, Brenowitz WD, Mayeda ER, et al. Rate of Memory Change Before and After Cancer Diagnosis. JAMA Netw Open 2019;2:e196160 10.1001/jamanetworkopen.2019.6160. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [28].Lanni C, Racchi M, Memo M, Govoni S, Uberti D. P53 at the Crossroads between Cancer and Neurodegeneration. Free Radic Biol Med 2012;52:1727–33. 10.1016/j.freeradbiomed.2012.02.034 [doi]. [DOI] [PubMed] [Google Scholar]
  • [29].Caricasole A, Bakker A, Copani A, Nicoletti F, Gaviraghi G, Terstappen GC. Two sides of the same coin: Wnt signaling in neurodegeneration and neuro-oncology. Biosci Rep 2005;25:309–27. 10.1007/s10540-005-2893-6 [doi]. [DOI] [PubMed] [Google Scholar]
  • [30].Entschladen F, Drell TL 4th, Lang K, Joseph J, Zaenker KS. Tumour-cell migration, invasion, and metastasis: navigation by neurotransmitters. The LancetOncology 2004;5:254–8. 10.1016/S1470-2045(04)01431-7 [doi]. [DOI] [PubMed] [Google Scholar]
  • [31].Hanoun M, Maryanovich M, Arnal-Estape A, Frenette PS. Neural regulation of hematopoiesis, inflammation, and cancer. Neuron 2015;86:360–73. 10.1016/j.neuron.2015.01.026 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [32].Ward BK, Mark PJ, Ingram DM, Minchin RF, Ratajczak T. Expression of the estrogen receptor-associated immunophilins, cyclophilin 40 and FKBP52, in breast cancer. Breast Cancer Res Treat 1999;58:267–80. [DOI] [PubMed] [Google Scholar]
  • [33].Baker JD, Shelton LB, Zheng D, Favretto F, Nordhues BA, Darling A, et al. Human cyclophilin 40 unravels neurotoxic amyloids. PLoS Biol 2017;15:e2001336 10.1371/journal.pbio.2001336 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [34].Chen Y, Cress RD, Stewart SL, Semrad TJ, Harvey D, Tencredi DJ, et al. Mediating Effect of Postsurgical Chemotherapy on Presence of Dementia and Survival among Patients 65 and Older with Stage III Colon Cancer. Cancer Epidemiol Biomarkers Prev 2017;26:1558–1563. 10.1158/1055-9965.EPI-17-0277 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [35].Driver JA, Djousse L, Logroscino G, Gaziano JM, Kurth T. Incidence of cardiovascular disease and cancer in advanced age: prospective cohort study. BMJ 2008;337:a2467 10.1136/bmj.a2467 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [36].Iadecola C. The overlap between neurodegenerative and vascular factors in the pathogenesis of dementia. Acta Neuropathol 2010;120:287–96. 10.1007/s00401-010-0718-6 [doi]. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES