Abstract

Background

   Observational studies have shown that the correlation between
   neurodegenerative diseases and colorectal cancer (CRC) remains
   controversial. Therefore, this study aimed to verify the causal
   association between these two diseases.

Methods

   Mendelian randomization (MR) analysis was used to assess the causal
   relationships between five major neurodegenerative diseases and CRC.
   Multivariable MR (MVMR) analysis was conducted to assess the direct
   causal effect of neurodegenerative diseases on CRC. Colocalization and
   pathway enrichment analyses were conducted to further elucidate our
   results. Sensitivity analysis was conducted to assess the robustness of
   the results.

Results

   Genetically predicted Alzheimer's disease (AD) nominally increased CRC
   risk (OR = 1.0620, 95%CI = 1.0127–1.1136, P = 0.013). There was no
   causal effect of genetically predicted CRC on neurodegenerative
   diseases. Furthermore, we demonstrated that genetically predicted AD
   marginally increased colon cancer risk (OR = 1.1621,
   95%CI = 1.0267–1.3153, P = 0.017). Genetically predicted Lewy body
   dementia (LBD) had a significant causal effect on the increasing risk
   of colon cancer (IVW OR = 1.1779, 95%CI = 1.0694–1.2975, P = 0.001).
   MVMR indicated that effect of AD on colon cancer was driven by LBD,
   type 2 diabetes, body mass index, low-density lipoprotein cholesterol,
   high-density lipoprotein cholesterol, triglyceride, total cholesterol
   (TC), processed meat consumption, smoking, alcohol consumption, and
   educational attainment, whereas the effect of LBD on colon cancer was
   only influenced by TC. Colocalization and pathway enrichment analysis
   suggested that LBD and colon cancer possibly shared causal variants
   (nearby gene APOE), and ERBB4 signaling and lipid metabolism may
   mediate the causal association between LBD and colon cancer.
   Sensitivity analysis confirmed the reliability of our findings.

Conclusions

   Our study demonstrated that genetic vulnerabilities to AD nominally
   increased the overall risk of CRC and colon cancer. Genetically
   predicted LBD indicated an elevated risk of colon cancer, potentially
   linked to ERBB4 signaling and lipid metabolism.

   Keywords: Colorectal cancer, Neurodegenerative disease, Mendelian
   randomization, Alzheimer's disease, Lewy body dementia, Colocalization
   analysis

Highlights

     * •
       Genetic vulnerabilities to AD nominally increased the risk of
       overall CRC and colon cancer.
     * •
       Genetically predicted LBD had a significant causal effect on the
       increasing risk of colon cancer.
     * •
       AD’s effect on colon cancer was driven by the common risk factors,
       while LBD’s effect was only driven by blood lipid.
     * •
       ERBB4 signaling and lipid metabolism might mediate the causal
       association between LBD and colon cancer.

1. Introduction

   Neurodegenerative diseases and colorectal cancer (CRC) are chronic and
   progressive diseases, for which aging is a significant risk factor
   [[29]1]. The incidence of these diseases is gradually increasing with
   the acceleration of population aging, which has a considerable
   socioeconomic impact [[30]2,[31]3]. Therefore, greater attention to
   these two diseases is imperative. Dysfunction, slow progressive loss of
   neurons in the central nervous system, impairment of synaptic
   plasticity, progressive muscle atrophy or wasting, and proteinopathies,
   such as misfolded amyloid-β (Aβ) and tau in Alzheimer's disease (AD),
   α-synuclein in Parkinson's disease (PD), are all characteristics of
   neurodegenerative diseases. These pathological processes eventually
   impair memory, movement, and cognition [[32]4]. CRC is the second major
   cause of cancer-related deaths and the third most common cancer
   globally. Therefore, it is crucial to investigate the risk factors and
   pathophysiology of CRC for better prevention and treatment [[33]2].
   Similar to other cancers, CRC is associated with many pathological
   processes, including excess cell proliferation, growth factor receptor
   overactivation, resistance to cell death, induction of angiogenesis,
   metastasis, and immune evasion [[34]5].

   Neurodegenerative diseases and cancer exhibit opposing mechanisms,
   involving progressive cell death and uncontrolled cell proliferation.
   Observational studies have shown an inverse correlation between
   neurodegenerative diseases and CRC [[35]6,[36]7]. Further research has
   identified consistent pathophysiological mechanisms in these two
   diseases, including mitochondrial dysfunction, autophagy, DNA damage,
   and oxidative stress [[37]8]. In addition, there is increasing evidence
   that the brain-gut axis plays a crucial role in connecting neurological
   and gastrointestinal disorders. Therefore, the association between
   neurodegenerative diseases and CRC has shifted from a negative
   correlation in the past to a positive correlation between certain
   neurodegenerative diseases and CRC at present, certain observational
   studies support this standpoint [[38]9,[39]10]. Unfortunately, because
   neurodegenerative diseases usually persist for over a decade, the
   actual onset of neurodegeneration might precede clinical manifestations
   by many years. This makes it difficult for observational studies to
   determine the precise onset times. Furthermore, various measurement
   errors, potential biases, and confounding factors (e.g., disease
   development and treatment) in the observational studies may have
   obscured the actual causal association.

   Mendelian randomization (MR) analysis can eliminate this bias in
   observational studies. Genetic variation is used as an instrument
   variable (IV) to deduce the causal association between exposure and
   outcome in the MR analysis. First, confounding factors can
   theoretically be avoided in MR analysis. Second, exposure is associated
   with IV, which appears before the outcome, thus excluding the influence
   of reverse causation [[40]11].

   In this study, we performed MR analysis to evaluate the causal
   relationship between five major neurodegenerative diseases, including
   AD, PD, amyotrophic lateral sclerosis (ALS), Lewy body dementia (LBD),
   frontotemporal dementia (FTD), and CRC (including colon cancer and
   rectum cancer). Multivariable MR (MVMR) analysis was conducted to
   assess the direct causal effect of neurodegenerative diseases on CRC.
   Finally, colocalization and pathway enrichment analyses were performed
   to further elucidate the results.

2. Materials and Methods

2.1. Study design

   We displayed the overview of our study in [41]Fig. 1. In general, there
   are three assumptions needed to be satisfied in the bidirectional MR
   analysis: (1) IVs for AD, PD, ALS, LBD, FTD, and CRC are closely linked
   to exposure; (2) IVs are not related to confounders in the
   exposure-outcome association; (3) IVs directly through exposure
   influence the outcome instead of through other pathways.

Fig. 1.

   [42]Fig. 1
   [43]Open in a new tab

   Overview of our study design. AD, Alzheimer's disease; PD, Parkinson's
   disease; ALS, amyotrophic lateral sclerosis; LBD, Lewy body dementia;
   FTD, frontotemporal dementia; CRC, colorectal cancer.

2.2. Data sources

   Genetic association data of AD (case = 85,934, controls = 401,577,
   N = 487,511), PD (case = 33,674, controls = 449,056, N = 482,730), ALS
   (case = 27,205, controls = 110,881, N = 138,086), LBD (case = 2,591,
   controls = 4,027, N = 6618), FTD (case = 2,154, controls = 4,308,
   N = 6462), CRC (case = 78,473, controls = 107,143, N = 185,616), body
   mass index (BMI) (N = 681,275), smoking (cigarettes per day,
   N = 377,334), alcohol consumption (drinks per week, N = 941,280), and
   educational attainment (N = 405,072) were obtained from large-scale
   genome-wide association studies (GWAS) [[44][12], [45][13], [46][14],
   [47][15], [48][16], [49][17], [50][18], [51][19], [52][20]]. Summary
   data of GWAS for colon cancer (case = 1,482, controls = 454,794,
   N = 456,276), rectum cancer (case = 326, controls = 455,950,
   N = 456,276), total cholesterol (TC) (N = 437,878), and processed meat
   consumption (N = 448,303) were from the UK Biobank study [[53]21].
   Genetic association data of low-density lipoprotein cholesterol (LDL-C)
   (N = 343,621), high-density lipoprotein cholesterol (HDL-C)
   (N = 315,133), triglyceride (TG) (N = 343,992), and type 2 diabetes
   (T2D) (N = 462,933) were from the IEU Open GWAS database
   ([54]https://gwas.mrcieu.ac.uk/). The detailed data sources are listed
   in [55]Table S1.

2.3. Instrumental variables selection

   We confirmed single nucleotide polymorphisms (SNPs) that were
   associated with genome-wide significance (P < 5 × 10^−8) associations
   for exposure and performed linkage disequilibrium (LD) clumping by
   setting r^2＜0.001 and clump distance＞10,000 kb to ensure the SNPs are
   independent and satisfy the first assumption. Because of only one IV
   for FTD, we used a more relaxed significance threshold (P < 1 × 10^−5).
   In addition, to prevent weak IVs, we calculated R^2
   (R^2 = 2 × EAF × (1−EAF) × β^2), representing the proportion of
   variation explained by IVs, and the F-statistic (F = R^2 (n-k-1)/k
   (1-R^2)), indicating the strong association between IVs and exposure.
   We identified 60 SNPs for AD, 23 for PD, 12 for ALS, 5 for LBD, 13 for
   FTD, and 116 for CRC. The details of IVs for exposures are shown in
   [56]Tables S2–S7.

2.4. Mendelian randomization analysis

   Four common MR approaches, including inverse variance weighted (IVW),
   weighted median (WM), MR-Egger, and MR-Pleiotropy Residual Sum and
   Outlier (MR-PRESSO) methods, were utilized to evaluate causal
   associations between five neurodegenerative diseases and CRC, including
   colon cancer and rectum cancer. IVW is the main evaluation method
   because it has the most significant statistical power and can identify
   reliable causal estimates without directional pleiotropy [[57]22].
   Because MR-Egger and WM methods could provide more robust but wider
   confidence interval estimations in broader conditions, we used them to
   complement IVW estimations.

   To eliminate the influence of confounders, we used two different
   methods. In the first one, we searched for pleiotropic SNPs in
   PhenoScanner ([58]http://www.phenoscanner.medschl.cam.ac.uk/). Then, we
   excluded IVs significantly associated (P < 1 × 10^−5) with other
   traits, including BMI, T2D, smoking, alcohol consumption, processed
   meat consumption, inflammatory bowel disease, and cholesterol, which
   could affect our results and used the remaining IVs for MR analysis. In
   the second one, we conducted MVMR analysis with adjustment for
   genetically predicted other identified neurodegenerative diseases, T2D,
   BMI, LDL-C, HDL-C, TG, TC, processed meat consumption, smoking, alcohol
   consumption, and educational attainment, which have been demonstrated
   to be associated with CRC [[59][23], [60][24], [61][25], [62][26]].

   Furthermore, we conducted a comprehensive sensitivity analysis to
   detect potential violations of the model assumptions in the MR
   analysis. We used Cochran's Q test to evaluate the heterogeneity among
   IVs (value of P < 0.05 suggesting heterogeneity). The MR-PRESSO method
   was also utilized to detect outlier IVs and correct horizontal
   pleiotropy. We conducted the MR-Egger intercept test to detect the
   directional pleiotropy of IVs (value of P < 0.05 suggesting pleiotropy)
   [[63]27]. We also performed the Leave-one-out analysis to evaluate the
   reliability of the results of the MR analysis.

   All MR analyses were conducted by the “TwoSampleMR” R package, the
   “MendelianRandomization” R package and the “MRPRESSO” R package in R
   software (version 4.2.2). Multiple comparisons were made by Bonferroni
   correction (P < 0.05/5), which was regarded as evidence of statistical
   significance. However, P < 0.05 was assumed nominally significant of a
   causal association.

2.5. Colocalization and enrichment analysis

   For neurodegenerative diseases that appeared causally associated with
   CRC risk in the MR analysis, we performed colocalization analysis using
   the “coloc” R package to evaluate the probability of shared causal
   variants between them. Colocalization analysis was performed by
   generating ±50 kb windows from the top SNP of neurodegenerative
   diseases. We obtained posterior probability for 5 hypotheses (H0–H4) in
   a Bayesian framework [[64]28]. PP.H4 ≥ 80 % of the colocalization
   analysis (H4) indicated strong-evidence for shared causal variants.
   Medium-strength evidence for shared causal variants was defined as
   50 % < PP.H4 < 80 % [[65]29]. Finally, we conducted the pathway
   enrichment analysis to identify overrepresented pathways in the top SNP
   utilizing Reactome [[66]30]. The results were adjusted for false
   discovery rate.

3. Results

3.1. Effect of neurodegenerative diseases on CRC

   After excluding potential pleiotropic and palindromic SNPs, we used the
   remaining SNPs to evaluate the effect of neurodegenerative diseases on
   CRC. Based on Bonferroni correction, we found that genetically
   predicted AD could nominally raise the risk of CRC using the IVW method
   (IVW OR = 1.0620, 95%CI = 1.0127–1.1136, P = 0.013). Moreover, the
   consistent impact direction was shown in other methods ([67]Fig. 2).
   The MR-PRESSO global test showed no horizontal pleiotropy (P = 0.051).
   The MR analysis showed no effect of PD on CRC. However, the MR-PRESSO
   global test showed horizontal pleiotropy (P = 0.013). After one outlier
   was removed (rs6741007), the result of the IVW method was still the
   same as before (IVW P = 0.447) ([68]Fig. 2). There was no causal effect
   of genetically predicted ALS on CRC in all MR methods. However, the
   MR-PRESSO global test showed horizontal pleiotropy (P = 0.006) and
   found one outlier (rs517339). After the outlier was removed, the
   results of the MR analysis were not changed (IVW P = 0.814) ([69]Fig.
   2). There was no causal association between LBD and CRC in all MR
   methods ([70]Fig. 2). Furthermore, the MR-PRESSO global test showed no
   horizontal pleiotropy (P = 0.116). The MR analysis showed no effect of
   FTD on CRC. However, the MR-PRESSO global test showed horizontal
   pleiotropy (P < 0.001). After one outlier was removed (rs9268877), the
   result of the IVW method was not changed (IVW P = 0.364) ([71]Fig. 2).

Fig. 2.

   [72]Fig. 2
   [73]Open in a new tab

   Forest plot for results of the bidirectional Mendelian randomization
   analysis. AD, Alzheimer's disease; PD, Parkinson's disease; ALS,
   amyotrophic lateral sclerosis; LBD, Lewy body dementia; FTD,
   frontotemporal dementia; CRC, colorectal cancer.

3.2. Effect of CRC on neurodegenerative diseases

   After excluding potential pleiotropic and palindromic SNPs, we used the
   remaining SNPs to evaluate the effect of CRC on neurodegenerative
   diseases. We found no effect of CRC on AD using the MR analysis.
   However, the MR-PRESSO global test showed horizontal pleiotropy
   (P < 0.001). After two outliers were removed (rs2527927 and
   rs11692435), the results of MR analysis were consistent with before
   (IVW P = 0.206) ([74]Fig. 2). We did not find that a genetic
   predisposition to CRC was associated with PD. However, the MR-PRESSO
   global test showed horizontal pleiotropy (P = 0.010). After one outlier
   was removed (rs12078075), the results of the MR analysis were not
   changed (IVW P = 0.397, MR-Egger P = 0.339, WM P = 0.660, MR-PRESSO
   P = 0.399) ([75]Fig. 2). In addition, MR analysis showed no effect of
   CRC on ALS (IVW P = 0.665) ([76]Fig. 2). The MR-PRESSO global test
   showed no horizontal pleiotropy (P = 0.107). There was also no causal
   association between CRC and LBD in all MR methods ([77]Fig. 2). The
   MR-PRESSO global test showed no horizontal pleiotropy (P = 0.116). We
   found no effect of CRC on FTD using the MR analysis (IVW P = 0.160)
   ([78]Fig. 2). Moreover, the MR-PRESSO global test showed no horizontal
   pleiotropy (P = 0.740).

3.3. Sensitivity analysis

   We used several sensitivity analyses to evaluate the reliability of our
   MR analysis. The Cochran's Q test showed slight heterogeneity in
   partial MR estimates, and we used the random effects IVW method to
   alleviate the problem ([79]Table S8). The MR-Egger intercept test
   indicated no directional pleiotropy in all MR estimates ([80]Table S8).
   Forest and funnel plots among all IVs in the bidirectional MR analysis
   are shown in [81]Figs. S1–S5. The plots of the leave-one-out
   sensitivity analysis demonstrated that no SNP affected the overall
   results in our bidirectional MR analysis ([82]Figs. S1–S5).

3.4. Effect of neurodegenerative diseases on colon cancer and rectum cancer

   We further explored the causal association between neurodegenerative
   diseases and the site of CRC. After Bonferroni correction for multiple
   tests, we found that genetically predicted AD nominally increased the
   risk of colon cancer (IVW OR = 1.1621, 95%CI = 1.0267–1.3153,
   P = 0.017) ([83]Table 1). LBD and colon cancer showed a significant
   causal association. we found that genetically predicted LBD increased
   the risk of colon cancer (IVW OR = 1.1779, 95%CI = 1.0694–1.2975,
   P = 0.001) ([84]Table 1). However, we found no significant causal
   association between the other three neurodegenerative diseases and
   colon cancer ([85]Table 1). The Cochran's Q test indicated no
   heterogeneity, and the MR-Egger intercept and MR-PRESSO global tests
   also showed no significant horizontal pleiotropy in all MR estimates
   ([86]Table S9).

Table 1.

   Results of the MR analysis between neurodegenerative diseases and colon
   cancer and rectum cancer. AD, Alzheimer's disease; PD, Parkinson's
   disease; ALS, amyotrophic lateral sclerosis; LBD, Lewy body dementia;
   FTD, frontotemporal dementia.
   Exposure Outcome MR method OR (95 % CI) P
   AD Colon cancer Inverse variance weighted 1.1621 (1.0267–1.3153) 0.017
   MR-Egger 1.1913 (0.9599–1.4786) 0.118
   Weighted median 1.2439 (1.0312–1.5005) 0.023
   MR-PRESSO 1.1659 (1.0474–1.2844) 0.011
   AD Rectum cancer Inverse variance weighted 0.7999 (0.6146–1.0409) 0.097
   MR-Egger 1.1591 (0.7322–1.8347) 0.531
   Weighted median 1.0890 (0.7450–1.5917) 0.660
   MR-PRESSO 0.8126 (0.5627–1.0626) 0.104
   PD Colon cancer Inverse variance weighted 0.9242 (0.8230–1.0379) 0.183
   MR-Egger 1.1043 (0.8233–1.4813) 0.515
   Weighted median 0.9066 (0.7722–1.0644) 0.231
   MR-PRESSO 0.9242 (0.8201–1.0283) 0.152
   PD Rectum cancer Inverse variance weighted 1.0219 (0.9097–1.1480) 0.715
   MR-Egger 1.0284 (0.7662–1.3803) 0.854
   Weighted median 0.8140 (0.9618–1.1364) 0.647
   MR-PRESSO 1.0219 (0.9183–1.1255) 0.686
   ALS Colon cancer Inverse variance weighted 0.9247 (0.7271–1.1760) 0.523
   MR-Egger 0.9959 (0.5588–1.7752) 0.989
   Weighted median 1.0932 (0.7764–1.5394) 0.610
   MR-PRESSO 0.9247 (0.6870–1.1624) 0.532
   ALS Rectum cancer Inverse variance weighted 1.1127 (0.8743–1.4162)
   0.385
   MR-Egger 1.3505 (0.7706–2.3670) 0.319
   Weighted median 1.1992 (0.8710–1.6510) 0.266
   MR-PRESSO 1.1127 (0.8890–1.3365) 0.471
   LBD Colon cancer Inverse variance weighted 1.1779 (1.0694–1.2975) 0.001
   MR-Egger 1.2499 (1.0630–1.4696) 0.007
   Weighted median 1.2051 (1.0794–1.3456) 0.001
   MR-PRESSO 1.1779 (1.1089–1.2469) 0.010
   LBD Rectum cancer Inverse variance weighted 1.0978 (0.9965–1.2095)
   0.059
   MR-Egger 1.2304 (1.0462–1.4471) 0.009
   Weighted median 1.1022 (0.9879–1.2298) 0.081
   MR-PRESSO 1.0978 (1.0066–1.1891) 0.012
   FTD Colon cancer Inverse variance weighted 0.9322 (0.8349–1.0409) 0.212
   MR-Egger 0.8563 (0.5985–1.2252) 0.414
   Weighted median 0.9119 (0.8039–1.0344) 0.151
   MR-PRESSO 0.9322 (0.8220–1.0425) 0.634
   FTD Rectum cancer Inverse variance weighted 0.9783 (0.8759–1.0928)
   0.698
   MR-Egger 0.8076 (0.5732–1.1381) 0.248
   Weighted median 1.0097 (0.8802–1.1583) 0.890
   MR-PRESSO 0.9784 (0.8677–1.0890) 0.705
   [87]Open in a new tab

   The results of the MR analysis showed no significant causal association
   between neurodegenerative diseases and rectum cancer ([88]Table 1). The
   Cochran's Q test indicated no heterogeneity, and the MR-Egger intercept
   and MR-PRESSO global tests also showed no significant horizontal
   pleiotropy in all MR estimates ([89]Table S9).

   [90]Figs. S6–10 show the forest and funnel plots among all IVs in the
   MR analysis. The plots of the leave-one-out sensitivity analysis
   demonstrated that our findings were reliable ([91]Figs. S6–10).

3.5. MVMR analysis

   The MVMR analysis was conducted to assess whether the association
   between genetically predicted AD and LBD and increased risk of colon
   cancer was influenced by other neurodegenerative diseases, T2D, BMI,
   LDL-C, HDL-C, TG, TC, processed meat consumption, smoking, alcohol
   consumption, and educational attainment. When adjusting for these
   factors, the effect of genetically predicted AD on colon cancer was
   attenuated ([92]Table 2). However, the association between genetically
   predicted LBD and increased risk of colon cancer was only driven by TC
   ([93]Table 3). The MVMR Egger intercept test showed no horizontal
   pleiotropy ([94]Table 2, [95]Table 3).

Table 2.

   Results of the MVMR analysis between AD and colon cancer. AD,
   Alzheimer's disease; LBD, Lewy body dementia; T2D, type 2 diabetes;
   BMI, body mass index; LDL, low-density lipoprotein cholesterol; HDL,
   high-density lipoprotein cholesterol; TG, triglyceride; TC, total
   cholesterol.
   Adjustment OR (95 % CI) P MVMR Egger
     __________________________________________________________________

   Intercept SE P
   LBD 0.9436 (0.8123–1.0750) 0.385 −0.011 0.008 0.200
   T2D 0.8869 (0.7184–1.0555) 0.164 −0.009 0.018 0.619
   BMI 0.9185 (0.6598–1.1772) 0.521 0.001 0.003 0.768
   LDL 0.9792 (0.8361–1.1223) 0.773 −0.008 0.005 0.112
   HDL 1.0346 (0.8954–1.1737) 0.628 0.001 0.003 0.765
   TG 1.1618 (0.9796–1.3441) 0.105 −0.005 0.006 0.439
   TC 0.9474 (0.6991–1.1958) 0.670 0.001 0.005 0.885
   Processed meat consumption 0.9901 (0.8613–1.1188) 0.879 −0.003 0.008
   0.691
   Smoking 0.9871 (0.8361–1.1381) 0.866 −0.005 0.009 0.593
   Alcohol consumption 0.9841 (0.8571–1.1112) 0.805 −0.008 0.007 0.222
   Educational attainment 0.9753 (0.8350–1.1157) 0.727 −0.006 0.005 0.261
   [96]Open in a new tab

Table 3.

   Results of the MVMR analysis between LBD and colon cancer. AD,
   Alzheimer's disease; LBD, Lewy body dementia; T2D, type 2 diabetes;
   BMI, body mass index; LDL, low-density lipoprotein cholesterol; HDL,
   high-density lipoprotein cholesterol; TG, triglyceride; TC, total
   cholesterol.
   Adjustment OR (95 % CI) P MVMR Egger
     __________________________________________________________________

   Intercept SE P
   AD 1.1366 (1.0248–1.2483) 0.024 −0.011 0.008 0.200
   T2D 1.1780 (1.0610–1.3080) 0.002 −0.034 0.028 0.231
   BMI 1.0754 (1.0007–1.1557) 0.048 0.001 0.005 0.865
   LDL 1.1140 (1.0102–1.2179) 0.042 0.000 0.006 0.979
   HDL 1.0931 (1.0127–1.1734) 0.029 −0.004 0.004 0.305
   TG 1.1514 (1.0730–1.2298) 0.000 0.003 0.004 0.388
   TC 1.0030 (0.8609–1.1451) 0.967 −0.006 0.005 0.267
   Processed meat consumption 1.1901 (1.0940–1.2861) 0.000 0.010 0.012
   0.398
   Smoking 1.1651 (1.0544–1.2874) 0.003 0.008 0.015 0.610
   Alcohol consumption 1.1607 (1.0627–1.2587) 0.003 −0.012 0.009 0.183
   Educational attainment 1.1677 (1.0775–1.2578) 0.001 −0.021 0.019 0.251
   [97]Open in a new tab

3.6. Colocalization and pathway enrichment analysis

   We performed colocalization analysis to evaluate the probability of
   shared causal variants between neurodegenerative diseases (AD and LBD)
   and colon cancer. There was no evidence that AD and colon cancer shared
   causal variants (PP.H4 = 8.1 %) ([98]Fig. S11). However, there was
   medium-strength evidence that LBD and colon cancer shared causal
   variants (PP.H4 = 61.0 %) ([99]Fig. 3B). Regional Manhattan plots
   examined the association of all SNPs ±50 kb from the top SNP for LBD
   (rs769449, nearby gene APOE) for their association with LBD and with
   colon cancer risk ([100]Fig. 3A). Then, we performed pathway enrichment
   analysis for APOE. The results of enrichment analysis indicated that
   twenty-eight out of thirty-four pathways were significantly enriched
   ([101]Table S10). They are closely associated with ERBB4 signaling,
   NR1H2 and NR1H3-mediated signaling, and lipid metabolism ([102]Fig. 4
   and [103]Table S10). It can be inferred that these pathways may mediate
   the causal association between LBD and colon cancer. [104]Fig. 4
   displayed the top 20 significantly enriched pathways in the pathway
   enrichment analysis.

Fig. 3.

   [105]Fig. 3
   [106]Open in a new tab

   (A) Regional association plots for LBD and colon cancer at the
   chromosome 19 locus overlapping APOE. (B) Colocalization analysis of
   LBD and colon cancer. PP.H0 = neither LBD nor colon cancer risk has a
   genetic association in the region, PP.H1 = only LBD has a genetic
   association in the region, PP.H2 = only colon cancer risk has a genetic
   association in the region, PP.H3 = both LBD and colon cancer risk are
   associated but have different causal variants, PP.H4 = both LBD and
   colon cancer risk are associated and share a single causal variant.

Fig. 4.

   [107]Fig. 4
   [108]Open in a new tab

   Top 20 significantly enriched pathways in the pathway enrichment
   analysis.

4. Discussion

   Using MR analysis, our study provided convincing evidence that genetic
   vulnerabilities to AD nominally increased the risk of CRC (IVW
   OR = 1.0620, 95%CI = 1.0127–1.1136, P = 0.013). However, there was no
   causal association between the other four neurodegenerative diseases
   (PD, ALS, LBD, and FTD) and CRC. We found that CRC had no causal effect
   on neurodegenerative diseases. Furthermore, genetically predicted AD
   nominally increased the risk of colon cancer (IVW OR = 1.1621,
   95%CI = 1.0267–1.3153, P = 0.017). Genetically predicted LBD
   demonstrated a significant causal effect on the increased risk of colon
   cancer (IVW OR = 1.1779, 95%CI = 1.0694–1.2975, P = 0.001). MVMR
   analysis confirmed that the effect of AD on colon cancer was partially
   influenced by LBD, T2D, BMI, LDL-C, HDL-C, TG, TC, processed meat
   consumption, smoking, alcohol consumption, and educational attainment.
   However, the association between genetically predicted LBD and an
   increased risk of colon cancer was only driven by TC. Colocalization
   analysis suggested that AD and colon cancer shared no causal variants
   (PP.H4 = 8.1 %), whereas LBD and colon cancer possibly shared causal
   variants (PP.H4 = 61.0 %) and had one shared locus (rs769449, nearby
   gene APOE). ERBB4 signaling and lipid metabolism may mediate the causal
   association between LBD and colon cancer. To the best of our knowledge,
   this is the first study to construct an integrative analysis to explore
   causal associations between neurodegenerative diseases and CRC.

   Given the increasing incidences of neurodegenerative diseases and CRC,
   it is necessary to elucidate the relationship between these two
   diseases. However, the association between these two diseases remains
   controversial according to current observational studies. MR analysis
   is not susceptible to confounding factors and reverse causality bias,
   allowing for more accurate conclusions. We found that genetic
   vulnerabilities to AD nominally increased the risk of CRC, which is
   consistent with previous observational research. An observational study
   revealed that individuals at high risk for AD had a considerably higher
   incidence of CRC than other populations (RR = 1.30, 95%CI = 1.21–1.39,
   P < 0.001) [[109]10]. It should be noted that the findings of previous
   studies differ from ours. A cohort study reported contradictory
   findings, revealing an inverse correlation between AD and CRC [[110]6].
   The findings may have resulted from the interaction between genetics
   and environment, where the non-genetic effect might have had a negative
   correlation and a greater magnitude than the positively correlated
   genetic effect. The actual sequence and time of onset of AD and CRC
   were not available in the observational study because of the long
   incubation periods of AD and CRC. These factors could be crucial causes
   for the inverse correlation between AD and CRC risk observed in the
   aforementioned study. Previous observational studies have revealed that
   patients with PD and ALS may have a decreased incidence of CRC;
   nevertheless, no study has explored the association between LBD and CRC
   [[111]7]. However, our study indicated that the associations observed
   in previous studies may be related to other factors, including racial
   differences, confounders, or biases, rather than genetic variants. We
   found no causal relationship between the other four neurodegenerative
   diseases (PD, ALS, LBD, and FTD) and CRC, possibly because of
   differences in the pathogenesis of neurodegenerative diseases. However,
   the potential mechanisms underlying the association between these
   neurodegenerative diseases and CRC remain unclear. Moreover, we found
   no causal effect of genetically predicted CRC on neurodegenerative
   diseases.

   Furthermore, we observed that genetically predicted AD nominally
   increased the risk for colon cancer. The MVMR analysis further
   confirmed that the effect of AD on colon cancer was influenced by
   common risk factors of colon cancer. Colocalization analysis suggested
   that AD and colon cancer shared no causal variants. The main
   characteristics of AD pathology are abnormal deposits of Aβ protein and
   phosphorylated tau in the patients with AD. AD model mice displayed the
   accumulation of Aβ in enteric neurons, which could lead to neuronal
   structure damage and a decrease in neuronal numbers. This impairment of
   intestinal barrier function may contribute to tumorigenesis [[112]31].
   BECN1 is critical in regulating autophagy and maintaining cellular
   homeostasis. The expression of BECN1 was significantly decreased in the
   brains of patients with AD [[113]32]. Low expression of BECN1, a tumor
   suppressor, can lead to tumorigenesis [[114]33]. A previous
   meta-analysis of the gut microbiome dataset of AD patients revealed
   alterations of significantly lower abundance of short-chain fatty acids
   producers, including Firmicutes, Clostridiaceae, Lachnospiraceae, and
   Rikenellaceae [[115]34]. The changes in gut microbiota and their
   Metabolites might promote gut inflammation and injure the intestinal
   barrier, increasing the risk of colon cancer [[116]35]. Although we
   found that genetic vulnerabilities to AD could increase the risk of
   colon cancer, MVMR and colocalization analysis might not support this
   result. Furthermore, it was undeniable that there were some opposite
   mechanisms between these two diseases. Research demonstrated that
   inhibition of Pin 1 and the Wnt signaling pathway, together with
   upregulation of p53, were found in the development of AD. However,
   these mechanisms were completely reversed in colon cancer [[117]36].
   Therefore, vivo and vitro studies on this topic are still necessary for
   the future. In general, the underlying mechanisms between AD and CRC
   are complex and deserve further investigation.

   We found that genetically predicted LBD predicted a higher risk of
   colon cancer. MVMR analysis indicated that the association between
   genetically predicted LBD and increased risk of colon cancer was only
   driven by TC. Moreover, colocalization analysis suggested that LBD and
   colon cancer possibly shared causal variants and had one shared locus
   (rs769449, nearby gene APOE). Enrichment analysis indicated that ERBB4
   signaling and lipid metabolism might mediate the causal association
   between LBD and colon cancer. The main pathogenesis of LBD is the
   deposition of α-synuclein in the nervous system and brain [[118]37]. We
   first reported that genetically predicted LBD could increase the risk
   of colon cancer. Because there is an overlap in the genetics of LBD and
   AD, we presume that they may increase the risk of colon cancer through
   the same mechanism [[119]38]. Both two diseases may lead to the
   accumulation of misfolded proteins in the gut, which can promote
   intestinal inflammation and tumorigenesis. Previous studies reported
   that Biological aggression and a poor prognosis in CRC were linked to
   APOE overexpression [[120]39]. Moreover, APOE is demonstrated to be a
   strong genetic risk factor for AD and LBD [[121]40]. Our study showed
   that APOE played a major role in ERBB4 signaling and lipid metabolism.
   Dysregulated lipid metabolism is demonstrated to be a risk factor for
   CRC and LBD [[122]40,[123]41]. Several malignancies, including CRC, are
   associated with the mutation or increased expression of ERBB4
   [[124]42]. ERBB4 promotes the progression of CRC by promoting
   epithelial-mesenchymal transition [[125]43]. Research showed that ERBB4
   played an essential role in AD [[126]44]. However, research on the
   association between ERBB4 and LBD is limited. Our findings indicated
   that ERBB4 signaling and lipid metabolism seemed likely to play a role
   in the LBD-colon cancer pathway. We provided a new perspective on the
   association between LBD and colon cancer. Further research is needed to
   explore potential underlying mechanisms between these two diseases.

   There are several major strengths in the study. First, the main
   strength was the MR design, which was based on genetic association
   data, and the results were not influenced by confounding factors.
   Second, we provided an explanation of the possible mechanisms
   underlying these relationships together with previous research, which
   strengthened the reliability of our findings. Third, we demonstrated
   that AD and LBD had causal effects on CRC. Therefore, it was necessary
   to strengthen the prevention and screening of CRC in patients with
   neurodegenerative diseases. Fourth, we conducted a bidirectional MR
   analysis to examine the direction of the causal relations.

   However, our study has several limitations. First, the GWAS data we
   used for this research were collected from the European population. It
   means whether our results correspond in other populations remained
   confirmed. Second, although the GWAS data of neurodegenerative diseases
   and CRC came from different samples in our study, it is inevitable that
   there may be a sample overlap. Third, we used a more relaxed
   significance threshold between IVs and FTD for obtaining more SNPs,
   which probably increased the risk of violating the first assumption of
   MR analysis. However, the F-statistic of each SNP was over 10,
   indicating that every IV included was robust. Fourth, we found no
   causal association between the other neurodegenerative diseases (PD,
   ALS, and FTD) and CRC, including colon cancer and rectum cancer.
   However, we cannot exclude the possibility that the association was not
   detected due to insufficient sample size at present. The association
   between neurodegenerative diseases and CRC might become significant,
   along with summary data from larger sample sizes of GWAS in the future.

5. Conclusion

   In summary, our study demonstrated that genetic vulnerabilities to AD
   nominally increased the overall risk of CRC and colon cancer.
   Genetically predicted LBD indicated an elevated risk of colon cancer.
   ERBB4 signaling and lipid metabolism may mediate the causal association
   between LBD and colon cancer. These findings demonstrated
   neurodegenerative diseases’ effect on CRC, which deserves further
   investigation.

Funding

   This research received no external funding.

Competing interests

   The authors declare no conflict of interest.

Ethical approval

   Review and/or approval by an ethics committee was not needed for this
   study because this study used the data from publicly available
   databases.

Informed consent statement

   Not applicable.

Consent to participate

   Not applicable.

Consent for publication

   Not applicable.

Data availability statement

   Only publicly available data were used in this study; the data sources
   and processing of these data are described in the Materials and
   Methods. Further inquiries can be directed to the corresponding author.

CRediT authorship contribution statement

   Feifan Wang: Writing – original draft, Software, Methodology,
   Conceptualization. Lu Chen: Validation, Methodology. Mengke Nie:
   Visualization, Validation. Zhongxin Li: Writing – review & editing,
   Supervision, Funding acquisition, Conceptualization.

Declaration of competing interest

   The authors declare that they have no known competing financial
   interests or personal relationships that could have appeared to
   influence the work reported in this paper.

Acknowledgements