First-onset mental disorders after cancer diagnosis and cancer-specific mortality: a nationwide cohort study

  • J. Zhu
    Correspondence
    Correspondence to: Dr Jianwei Zhu, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Box 281, Stockholm 171 77, Sweden. Tel: +46-762285864;
    Affiliations
    Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm;
    Search for articles by this author
  • F. Fang
    Affiliations
    Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm;
    Search for articles by this author
  • A. Sjölander
    Affiliations
    Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm;
    Search for articles by this author
  • K. Fall
    Affiliations
    Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm;

    Department of Clinical Epidemiology and Biostatistics, School of Medical Sciences, Örebro University, Örebro, Sweden;
    Search for articles by this author
  • H.O. Adami
    Affiliations
    Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm;

    Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, USA;

    Clinical Effectiveness Research Group, Institute of Health and Society, University of Oslo, Oslo, Norway;
    Search for articles by this author
  • U. Valdimarsdóttir
    Affiliations
    Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm;

    Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, USA;

    Center of Public Health Sciences, School of Health Sciences, Faculty of Medicine, University of Iceland, Reykjavík, Iceland
    Search for articles by this author

      ABSTRACT

      Background

      The diagnosis of cancer is strongly associated with the risk of mental disorders even in patients with no previous history of mental disorders. Accumulating data suggest that mental distress may accelerate tumor progression. We hypothesized therefore that mental disorders after a cancer diagnosis may increase the risk of cancer-specific mortality.

      Patients and methods

      We conducted a nationwide cohort study including 244 261 cancer patients diagnosed in Sweden during 2004–2009 and followed them through 2010. Through the Swedish Patient Register, we obtained clinical diagnoses of all mental disorders and focused on mood-, anxiety-, and substance abuse disorders (ICD10: F10–F16, F18–F19, F32–F33, F40–F41, and F43–45) that are commonly diagnosed among patients with cancer. We further classified the studied mental disorders into first-onset or recurrent mental disorders. We used Cox regression to estimate multivariable hazard ratios (HRs) with 95% confidence intervals (CIs) as a measure of the association between mental disorders after cancer diagnosis and cancer-specific mortality, adjusting for age, sex, calendar period, educational level, cancer stage, and cancer type at diagnosis.

      Results

      After cancer diagnosis, 11 457 patients were diagnosed with mood-, anxiety-, and substance abuse disorders; of which 7236 were first-onset mental disorders. Patients with a first-onset mental disorder were at increased risk of cancer-specific mortality (HR: 1.82, 95% CI: 1.71–1.92) while patients with a recurrent mental disorder had much lower risk elevation (HR: 1.14, 95% CI: 1.05–1.24). The increased cancer-specific mortality by first-onset mental disorders was observed for almost all cancer sites/groups and the association was stronger for localized cancers (HR: 2.00, 95% CI: 1.73–2.31) than for advanced cancers (HR: 1.49, 95% CI: 1.32–1.69).

      Conclusions

      Patients with a first-onset common mood-, anxiety-, or substance abuse disorder after cancer diagnosis may be at increased risk of cancer-specific death.

      Key words

      Introduction

      Mental distress is common among cancer patients; around one-third of cancer patients have been reported to have mood disorders in the hospital setting [
      • Mitchell A.J.
      • Chan M.
      • Bhatti H.
      • et al.
      Prevalence of depression, anxiety, and adjustment disorder in oncological, haematological, and palliative-care settings: a meta-analysis of 94 interview-based studies.
      ]. Beyond the burden of living with cancer and its treatments, receiving a cancer diagnosis is also a severely stressful event. An excessive or prolonged psychological reaction to the diagnosis may itself be associated with other serious health consequences, including various psychiatric disorders [
      • Mitchell A.J.
      • Chan M.
      • Bhatti H.
      • et al.
      Prevalence of depression, anxiety, and adjustment disorder in oncological, haematological, and palliative-care settings: a meta-analysis of 94 interview-based studies.
      ,
      • Vin-Raviv N.
      • Hillyer G.C.
      • Hershman D.L.
      • et al.
      Racial disparities in posttraumatic stress after diagnosis of localized breast cancer: the BQUAL study.
      ] and, in severe cases, even suicide and cardiovascular fatalities shortly after cancer diagnosis [
      • Fang F.
      • Fall K.
      • Mittleman M.A.
      • et al.
      Suicide and cardiovascular death after a cancer diagnosis.
      ].
      During the past decade, the evidence for the role of mental disorders in tumor progression and cancer-related mortality has been rapidly accumulating [
      • Chida Y.
      • Hamer M.
      • Wardle J.
      • Steptoe A.
      Do stress-related psychosocial factors contribute to cancer incidence and survival?.
      ]. Studies have reported various mental disorders to be associated with higher rate of all-cause [
      • Prasad S.M.
      • Eggener S.E.
      • Lipsitz S.R.
      • et al.
      Effect of depression on diagnosis, treatment, and mortality of men with clinically localized prostate cancer.
      • Watson M.
      • Haviland J.S.
      • Greer S.
      • et al.
      Influence of psychological response on survival in breast cancer: a population-based cohort study.
      • Wikman A.
      • Ljung R.
      • Johar A.
      • et al.
      Psychiatric morbidity and survival after surgery for esophageal cancer: a population-based cohort study.
      ] or cancer-specific [
      • Batty G.D.
      • Russ T.C.
      • Stamatakis E.
      • Kivimaki M.
      Psychological distress in relation to site specific cancer mortality: pooling of unpublished data from 16 prospective cohort studies.
      ,
      • Epplein M.
      • Zheng Y.
      • Zheng W.
      • et al.
      Quality of life after breast cancer diagnosis and survival.
      ] mortality. However, the findings of previous studies have often been inconsistent due to an array of methodological difficulties, including small sample sizes and self-reported symptoms. Large studies on clinically confirmed mental disorders, specifically those after cancer diagnosis, and their potential influence on cancer-specific mortality are lacking. Furthermore, few past studies have addressed the potentially different roles of first-onset and recurrent mental disorders on cancer-specific survival.
      Recently, we demonstrated a rapid rise in first-onset mood-, anxiety-, and substance abuse disorders in patients undergoing diagnostic work-up and immediately after cancer [
      • Lu D.
      • Andersson T.M.
      • Fall K.
      • et al.
      Clinical diagnosis of mental disorders immediately before and after cancer diagnosis: a Nationwide Matched Cohort Study in Sweden.
      ]. In the present study, we aimed to examine the role of these common mental disorders diagnosed after the cancer on cancer-specific mortality. As an indication of a severe stress-related reaction after cancer, we hypothesized specifically that such first-onset mental disorders would be associated with increased cancer-specific death.

      Materials and methods

       Study design

      In Sweden, reporting cancers has been required legally since 1958 and the completeness of Cancer Register approaches 100% [
      • Barlow L.
      • Westergren K.
      • Holmberg L.
      • Talback M.
      The completeness of the Swedish Cancer Register: a sample survey for year 1998.
      ]. We conducted a cohort study on all adult patients (age ≥30) diagnosed with a primary cancer between 1 January 2004 and 31 December 2009. After exclusion of diagnoses confirmed at autopsy, our analytic cohort comprised 244 261 cancer patients with information on age, sex, date of diagnosis, cancer stage, and the national registration number (NRN). The most common cancer types were prostate (N = 51 065), breast (N = 33 999), colorectal (N = 28 007), lung (N = 16 872), hematological (N = 15 907), and renal/bladder (N = 15 594) cancers, as well as malignant melanoma (N = 11 014). Additionally, highly fatal cancers of the esophagus, liver, and pancreas were pooled together as severe cancers (N = 6603). The individually unique NRN, allowed record linkages between the study cohort and the nationwide Patient, Causes of Death, Migration, Multi-Generation and Education Registers.

       Ascertainment of mental disorders

      In Sweden, all hospital discharge records have been collected in Patient Register since 1987 and >80% of outpatient visits for hospital-based specialist care since 2001 [
      • Ludvigsson J.F.
      • Andersson E.
      • Ekbom A.
      • et al.
      External review and validation of the Swedish national inpatient register.
      ]. In this register, all diagnoses are coded from 1997 onward according to the 10th Swedish revision of International Classification of Diseases (ICD). We focused on common mood-, anxiety-, and substance abuse disorders, which have been shown to have a dramatic risk elevation after cancer [
      • Lu D.
      • Andersson T.M.
      • Fall K.
      • et al.
      Clinical diagnosis of mental disorders immediately before and after cancer diagnosis: a Nationwide Matched Cohort Study in Sweden.
      ], and are potentially related to severe psychological stress [
      • Klengel T.
      • Binder EB.
      Epigenetics of Stress-related psychiatric disorders and gene x environment interactions.
      ]. These included substance abuse (ICD10: F10–F16/F18–F19), depression (ICD10: F32–F33), stress reaction/adjustment disorder (ICD10: F43), anxiety (ICD10: F40–F41), and somatoform/conversion disorder (ICD10: F44–F45). Other mental disorders (ICD10: F00–F99 excluding the studied diagnoses above) were classified as the secondary exposure, leaving cancer patients without any mental disorder after cancer as the unexposed group. The first diagnosis of mental disorders after the date of cancer diagnosis, either through inpatient or outpatient care was used as a time-varying variable, and patients were classified as exposed from the diagnosis date of mental disorder.

       Ascertainment of mortality

      Cancer patients were followed from the date of cancer diagnosis until death or 31 December 2010, whichever occurred first. From Causes of Death Register, we identified date and underlying cause of death. If the cancer diagnosis and the underlying cause of death indicated the same site/group of cancer, the study participant was defined as having a cancer-specific death (supplementary Table S1, available at Annals of Oncology online).

       Statistical analysis

      We used Cox regression to assess the association of mental disorders after cancer with cancer-specific mortality, estimating hazard ratios (HRs) with 95% confidence intervals (CIs).
      We first estimated the overall associations between the studied mental disorders and cancer-specific mortality, and then by mental disorders diagnosed within or beyond 90 days after cancer.
      We then carried out separate analyses for patients with and without a history of mental disorders before cancer. Patients diagnosed with mental disorders after cancer and without a previous history of any mental disorder before cancer (from 1 January 1987) were classified as having a first-onset mental disorder, while others were classified as having a recurrent mental disorder. The analysis was first conducted for all cancer types together and then separately for the most common sites/groups.
      To separately assess the role of specific mental disorders, we further carried out the analysis by individual diagnosis. To assess potential effect modifications, we further stratified the analysis by age, sex, calendar period, educational level, number of close relatives (including partner, children, and siblings; ≤2, 3–5, and ≥6), and cancer stage at diagnosis. The number of close relatives was ascertained through linking to Multi-Generation Register, which contains information on familial links for all Swedish residents born since 1932 onward.
      In all statistical models, we used age at follow-up as the underlying timescale and adjusted for age, sex, calendar period (2004–2006/2007–2009), educational level (>9/≤9 years), and cancer stage at diagnosis. In the analysis of all cancers, we further adjusted for cancer site/group (supplementary Table S1, available at Annals of Oncology online). In the analysis of hematological malignancies, subtype (Hodgkin/non-Hodgkin lymphoma, myeloma, and leukemia) was further adjusted for. Information on cancer stage was ascertained by the TNM or FIGO records, including localized (T-localized/N0/M0, FIGO-0/I), local spread (T-advanced/N0/M0, FIGO-II), regional spread (N+/M0, FIGO-III), and advanced (M+, FIGO-IV) cancers [
      • Berrino B.
      In Condensed TNM for Coding the Extent of Disease..
      ].
      Cancer patients with mental disorders adjacent to their diagnosis may receive different treatment regimens, leading to potentially different mortality. To test this hypothesis, we compared the percentage of and waiting-time for surgical treatments among patients with prostate, lung, or colorectal cancers (where surgical treatment is commonly used as the primary treatment), according to their exposure to the studied mental disorders. We also repeated the main analyses after further adjustment for surgery among patients with these cancers.
      All the statistical analyses were carried out in SAS9.4 and Stata13.1. The study was approved by the Regional Ethics Review Board in Stockholm, Sweden.

      Results

      After cancer diagnosis, 11 457 patients experienced the studied mental disorder including 7236 with first-onset, and 10 688 patients experienced other mental disorders including 6661 with first-onset. A higher proportion of patients with the studied mental disorders was female and diagnosed as cancer below age 65 years (Table 1).
      Table 1Baseline characteristics of study participants in a cohort study of 244 261 cancer patients in Sweden, diagnosed from 2004 to 2009
      Without mental disorders after cancerStudied mental disorders after cancer diagnosis
      a Studied mental disorders included depression (ICD10: F32–F33), anxiety (ICD: F40–F41), stress reaction and adjustment disorder (ICD10: F43), mental and behavioral disorders due to psychoactive substance use (ICD10: F10–F16, F18–F19), and somatoform/conversion disorder (ICD10: F44–F45).
      Other mental disorders after cancer diagnosis
      b Other mental disorders included any mental disorders (ICD10: F00–F99), except for the studied mental disorders.
      N222 11611 45710 688
      Sex (%)
       Male53.4545.2653.01
       Female46.5554.7446.99
      Age at follow-up (%)
       ≤65, years41.5856.5727.21
       66–7528.6824.0423.96
       >7529.7419.3948.83
      Calendar period at diagnosis (%)
       2004–200649.2655.5056.51
       2007–200950.7444.5043.49
      Educational level
       >9 years59.0863.7951.32
       ≤9 years40.8036.0148.62
       Missing0.120.200.065
      Previous mental disorders (%)
       No92.9863.1662.32
       Yes7.0236.8437.68
      Cancer stage
      c Cancer stage at diagnosis was defined according to TNM or FIGO records and was not available for hematological and central nervous system malignancies. Patients with hematological and central nervous system malignancies were classified in the group entitled ‘Missing’.
      (%)
       Localized41.8741.8841.90
       Local spread11.8211.6815.43
       Regional spread10.6113.9910.86
       Advanced9.926.406.05
       Unknown16.9216.1416.89
       Missing8.869.928.87
      Close relatives
      d Close relatives include partner, children, and siblings who have not died at cancer diagnosis. People with whom cancer patients had biological children are classified as partner.
       ≤223.1122.3130.81
       3–546.8945.4343.43
       ≥623.4826.7615.15
       Unknown6.525.5010.61
      N, number.
      a Studied mental disorders included depression (ICD10: F32–F33), anxiety (ICD: F40–F41), stress reaction and adjustment disorder (ICD10: F43), mental and behavioral disorders due to psychoactive substance use (ICD10: F10–F16, F18–F19), and somatoform/conversion disorder (ICD10: F44–F45).
      b Other mental disorders included any mental disorders (ICD10: F00–F99), except for the studied mental disorders.
      c Cancer stage at diagnosis was defined according to TNM or FIGO records and was not available for hematological and central nervous system malignancies. Patients with hematological and central nervous system malignancies were classified in the group entitled ‘Missing’.
      d Close relatives include partner, children, and siblings who have not died at cancer diagnosis. People with whom cancer patients had biological children are classified as partner.
      Patients with the studied mental disorders after cancer had a 53% increased cancer-specific mortality compared with patients without any mental disorders. The association was considerably stronger for first-onset than recurrent mental disorders (Table 2). No statistically significant association was noted between recurrent mental disorders and cancer-specific mortality in the analyses of specific cancer sites/groups while first-onset mental disorders were associated with a higher risk of cancer-specific mortality among all common cancer types (Figure 1). The increased cancer-specific mortality was observed for both mental disorders experienced within and beyond 90 days after cancer (Table 2).
      Table 2Association of studied mental disorders and cancer-specific mortality, shown by time from cancer diagnosis, stratified by previous mental disorders
      Studied mental disorders
      a Studied mental disorders included depression (ICD10: F32–F33), anxiety (ICD: F40–F41), stress reaction and adjustment disorder (ICD10: F43), mental and behavioral disorders due to psychoactive substance use (ICD10: F10–F16, F18–F19), and somatoform/conversion disorder (ICD10: F44–F45).
      Studied mental disorders within 90 days after cancerStudied mental disorders over 90 days after cancer
      NHR (95% CI)
      b HR, hazard ratio; CI, confident interval; models adjusted for age at cancer diagnosis, sex, calendar period of cancer diagnosis, cancer type, cancer stage at diagnosis, educational, and history of mental disorder before cancer diagnosis; patients without any mental disorders after cancer diagnosis were used as the reference group.
      NHR (95% CI)NHR (95% CI)
      Overall114571.53 (1.46–1.60)29851.30 (1.21–1.40)84721.68 (1.58–1.78)
      Previous mental disorders
      c A previous history of any mental disorder (after 1 January 1987) (ICD10: F00–F99).
       No72361.82 (1.71–1.92)13131.52 (1.37–1.69)59231.97 (1.84–2.11)
       Yes42211.14 (1.05–1.24)16721.10 (0.98–1.22)25491.22 (1.08–1.38)
      N, number.
      a Studied mental disorders included depression (ICD10: F32–F33), anxiety (ICD: F40–F41), stress reaction and adjustment disorder (ICD10: F43), mental and behavioral disorders due to psychoactive substance use (ICD10: F10–F16, F18–F19), and somatoform/conversion disorder (ICD10: F44–F45).
      b HR, hazard ratio; CI, confident interval; models adjusted for age at cancer diagnosis, sex, calendar period of cancer diagnosis, cancer type, cancer stage at diagnosis, educational, and history of mental disorder before cancer diagnosis; patients without any mental disorders after cancer diagnosis were used as the reference group.
      c A previous history of any mental disorder (after 1 January 1987) (ICD10: F00–F99).
      Figure thumbnail gr1
      Figure 1Hazard ratios of cancer-specific mortality among patients with studied mental disorders after cancer diagnosis when compared with patients without any mental disorders after cancer diagnosis, further stratified with respect to previous mental disorders (ICD10: F00–F99).
      Patients diagnosed with other mental disorders also showed increased cancer-specific mortality, regardless of whether the mental disorders were first-onset or recurrent (supplementary Table S2 and Figure S1, available at Annals of Oncology online).
      In the stratified analysis, the excess cancer-specific mortality by first-onset mood-, anxiety-, and substance abuse disorders did not appear to differ largely between men and women; neither by age, calendar period, educational level, or number of close relatives (Table 3). However, the association was stronger among patients with a diagnosis of lower stage cancers.
      Table 3Association of new-onset of studied mental disorders
      a Studied mental disorders included depression (ICD10: F32–F33), anxiety (ICD: F40–F41), stress reaction and adjustment disorder (ICD10: F43), mental and behavioral disorders due to psychoactive substance use (ICD10: F10–F16, F18–F19), and somatoform/conversion disorder (ICD10: F44–F45).
      after cancer and cancer-specific mortality, stratified by background characteristics
      N (%)HR (95% CI)
      b HR, hazard ratio; CI, confident interval; models adjusted for age at cancer diagnosis, sex, calendar period of cancer diagnosis, cancer type, cancer stage at diagnosis, educational, and history of mental disorder before cancer diagnosis; patients without any mental disorders after cancer diagnosis were used as the reference group.
      Sex
       Male3082 (42.59)1.93 (1.77–2.10)
       Female4154 (57.41)1.71 (1.58–1.85)
      Age at follow-up, years
       ≤653906 (53.98)1.73 (1.59–1.89)
       66–751785 (24.67)1.99 (1.78–2.21)
       >751545 (21.35)1.77 (1.59–1.98)
      Calendar period at diagnosis
       2004–20064310 (59.56)1.78 (1.65–1.91)
       2007–20092926 (40.44)1.89 (1.72–2.07)
      Educational level
       >9 years4705 (65.02)1.75 (1.62–1.88)
       ≤9 years2515 (34.76)1.88 (1.72–2.06)
      Cancer stage
      c Cancer stage at diagnosis was defined according to TNM or FIGO records and was not available for hematological and central nervous system malignancies. Patients with hematological and central nervous system malignancies were classified in the group entitled ‘Missing’.
       Localized3020 (41.74)2.00 (1.73–2.31)
       Local spread826 (11.42)2.04 (1.75–2.37)
       Regional spread1041 (14.39)1.85 (1.65–2.07)
       Advanced471 (6.51)1.49 (1.32–1.69)
       Unknown1139 (15.74)1.65 (1.43–1.90)
      Close relatives
      d Close relatives include partner, children, and siblings who have not died at cancer diagnosis. People with whom cancer patients had biological children are classified as partner.
       ≤21605 (22.18)1.76 (1.57–1.98)
       3–53375 (46.64)1.84 (1.69–2.01)
       ≥61865 (25.77)1.84 (1.63–2.09)
       Unknown391 (5.40)1.71 (1.30–2.23)
      N, number.
      a Studied mental disorders included depression (ICD10: F32–F33), anxiety (ICD: F40–F41), stress reaction and adjustment disorder (ICD10: F43), mental and behavioral disorders due to psychoactive substance use (ICD10: F10–F16, F18–F19), and somatoform/conversion disorder (ICD10: F44–F45).
      b HR, hazard ratio; CI, confident interval; models adjusted for age at cancer diagnosis, sex, calendar period of cancer diagnosis, cancer type, cancer stage at diagnosis, educational, and history of mental disorder before cancer diagnosis; patients without any mental disorders after cancer diagnosis were used as the reference group.
      c Cancer stage at diagnosis was defined according to TNM or FIGO records and was not available for hematological and central nervous system malignancies. Patients with hematological and central nervous system malignancies were classified in the group entitled ‘Missing’.
      d Close relatives include partner, children, and siblings who have not died at cancer diagnosis. People with whom cancer patients had biological children are classified as partner.
      Almost all subtypes of the first-onset mental disorders (depression, anxiety, stress reaction/adjustment disorder, and substance abuse) were associated with increased cancer-specific mortality (Table 4).
      Table 4Association of specific diagnosis of new-onset studied mental disorders
      a Studied mental disorders included depression (ICD10: F32–F33), anxiety (ICD: F40–F41), stress reaction and adjustment disorder (ICD10: F43), mental and behavioral disorders due to psychoactive substance use (ICD10: F10–F16, F18–F19), and somatoform/conversion disorder (ICD10: F44–F45).
      after cancer and cancer-specific mortality, stratified by previous mental disorder
      b Any mental disorders (ICD10: F00–F99).
      N (%)HR (95% CI)
      c HR, hazard ratio; CI, confident interval; models adjusted for age at cancer diagnosis, sex, calendar period of cancer diagnosis, cancer type, cancer stage at diagnosis, educational, and history of mental disorder before cancer diagnosis; patients without any mental disorders after cancer diagnosis were used as the reference group.
      Stress reaction832 (11.50)1.78 (1.51–2.11)
      Depression3109 (42.97)1.76 (1.62–1.92)
      Anxiety2061 (28.48)2.11 (1.92–2.33)
      Substance abuse926 (12.80)1.50 (1.25–1.79)
      Somatoform/conversion disorder308 (4.26)1.20 (0.82–1.76)
      N, number.
      a Studied mental disorders included depression (ICD10: F32–F33), anxiety (ICD: F40–F41), stress reaction and adjustment disorder (ICD10: F43), mental and behavioral disorders due to psychoactive substance use (ICD10: F10–F16, F18–F19), and somatoform/conversion disorder (ICD10: F44–F45).
      b Any mental disorders (ICD10: F00–F99).
      c HR, hazard ratio; CI, confident interval; models adjusted for age at cancer diagnosis, sex, calendar period of cancer diagnosis, cancer type, cancer stage at diagnosis, educational, and history of mental disorder before cancer diagnosis; patients without any mental disorders after cancer diagnosis were used as the reference group.
      Compared with unexposed patients, patients with a first-onset mood-, anxiety-, or substance abuse disorder had a similar (prostate cancer, P > 0.05) or a slightly higher (lung or colorectal cancer, P < 0.05) prevalence of surgery (supplementary Table S3, available at Annals of Oncology online). Patients with these mental disorders had also similar waiting-time for surgery compared with unexposed patients (P > 0.05). Adding surgery to the original models did not alter the results.

      Discussion

      The findings from this nationwide cohort study of more than 240 000 individuals suggest that patients diagnosed for the first time with common mood-, anxiety-, and substance abuse disorders after their cancer experience increased risk of cancer-specific death. The increased cancer-specific mortality was noted across all tested cancer sites and was particularly strong in the case of early stage cancers. Our findings support the hypothesis that mental disorders may be strongly associated with survival prospects of patients diagnosed with cancer—a finding that calls for further exploration of mechanism as well as enhanced clinical monitoring and treatment of these symptoms among cancer patients.
      To the best of our knowledge, the present study is the first large scale endeavor to evaluate the influence of clinically confirmed first-onset mental disorders after cancer on cancer-specific mortality. Previous research lends support to an association between psychological disorders and cancer mortality [
      • Chida Y.
      • Hamer M.
      • Wardle J.
      • Steptoe A.
      Do stress-related psychosocial factors contribute to cancer incidence and survival?.
      ]. In line with earlier findings [
      • Batty G.D.
      • Russ T.C.
      • Stamatakis E.
      • Kivimaki M.
      Psychological distress in relation to site specific cancer mortality: pooling of unpublished data from 16 prospective cohort studies.
      ,
      • Epplein M.
      • Zheng Y.
      • Zheng W.
      • et al.
      Quality of life after breast cancer diagnosis and survival.
      ,
      • Pirl W.F.
      • Greer J.A.
      • Traeger L.
      • et al.
      Depression and survival in metastatic non-small-cell lung cancer: effects of early palliative care.
      ], our data suggest that cancer patients face shorter cancer-specific survival specifically when diagnosed with first-onset mood-, anxiety-, and substance abuse disorders after cancer. It is possible that first-onset mental disorders represent a severe stress reaction specifically due to the diagnosis and living with cancer whilst a recurrent onset may demark a general lability for mental disorders, which might play a smaller role in cancer progression. Similar findings were indeed observed in a recent Danish study, showing that only bereavement—with its associated risk of mental disorders—occurring after but not before cancer diagnosis was associated with cancer mortality [
      • Levav I.
      • Kohn R.
      • Iscovich J.
      • et al.
      Cancer incidence and survival following bereavement.
      ]. Nevertheless, why patients with recurrent mental disorders do not experience similar elevation in cancer-specific mortality is enigmatic and calls for further studies.
      Several different pathways may explain the associations between the studied mental disorders and cancer-specific mortality. First, a mental disorder occurring in conjunction with cancer may influence treatment decisions. Yet, our secondary analysis did not show less prevalence of or longer waiting-time for curative treatment (i.e. surgery) among patients with prostate, lung, and colorectal cancers that had first-onset mental disorders after cancer. Second, although adjuvant cancer treatments, e.g. chemotherapy or radiation, could increase the risk of mood disorders, they might also be associated with extended survival. Not taking into account adjuvant therapy could therefore result in an attenuation of our point estimates. Third, psychopharmacological treatment of mental disorders could potentially mediate the association between mental disorders and cancer survival. Data are still limited on the association of psychotic medication and cancer-specific mortality [
      • Ranjan A.
      • Gupta P.
      • Srivastava SK.
      Penfluridol: an antipsychotic agent suppresses metastatic tumor growth in triple-negative breast cancer by inhibiting integrin signaling axis.
      ,
      • Chubak J.
      • Buist D.S.
      • Boudreau D.M.
      • et al.
      Breast cancer recurrence risk in relation to antidepressant use after diagnosis.
      ]. Thus, the role of psycho-pharmacological treatment as a mechanism between first-onset mental disorders and cancer progression cannot be excluded and warrants further study.
      Finally, downstream physiologic consequences of the patients’ mental health may independently affect tumor progression and survival [
      • Reiche E.M.
      • Nunes S.O.
      • Morimoto HK.
      Stress, depression, the immune system, and cancer.
      ]. Mental disorders, e.g. anxiety and depression, can cause activation in the hypothalamic–pituitary–adrenal axis and autonomic nervous system with corresponding deregulation of circadian cortisol rhythm and release of catecholamine [
      • Reiche E.M.
      • Nunes S.O.
      • Morimoto HK.
      Stress, depression, the immune system, and cancer.
      ]. In an animal model where circadian cortisol rhythm was blunted, implanted osteosarcomas and pancreatic adenocarcinomas grew significantly faster than in sham-operated animals [
      • Filipski E.
      • King V.M.
      • Li X.
      • et al.
      Host circadian clock as a control point in tumor progression.
      ]. Furthermore, the loss of normal circadian variation in cortisol was associated with earlier mortality among lung [
      • Sephton S.E.
      • Lush E.
      • Dedert E.A.
      • et al.
      Diurnal cortisol rhythm as a predictor of lung cancer survival.
      ] and breast cancer [
      • Sephton S.E.
      • Sapolsky R.M.
      • Kraemer H.C.
      • Spiegel D.
      Diurnal cortisol rhythm as a predictor of breast cancer survival.
      ] patients.
      The major strength of our study is the large-scale population-based cohort design, the complete follow-up and the independently and prospectively collected data on cancer, mental disorders, and cause of death. A few limitations should nevertheless be noted. First, mental disorders are a group of diseases associated with multiple etiologies ranging from environmental to genetic factors. Our main exposure, first-onset mood-, anxiety-, and substance abuse disorders after cancer, may potentially be a proxy for stress reaction to receiving cancer diagnosis or living with cancer, but could also be related to other causes. Second, we used clinically confirmed mental disorders as exposure, which likely indicate the most severe cases of mental distress. Patients with milder forms of mental distress that did not receive a clinical diagnosis were classified as unexposed, probably leading to an underestimate of the real associations. Third, the burden of mental disorders among cancers of very poor prognosis might have been underestimated. Fourth, we did not have complete information on cancer treatments. However, our secondary analyses with further adjustment for surgical treatment did not alter the results of prostate, lung and colorectal cancers. Further, it is possible that small number of patients died from other comorbidities were misclassified as from cancer, or vice versa. However, it is unlikely that this misclassification was related to mental disorders and this source of error would only dilute the reported associations. Finally, in this study cancer patients were only followed for an average of 2.9 years. Thus, future studies with longer follow-up times are needed to explore the association between mental disorders and longer-term survival.
      In conclusion, our findings suggest that patients experiencing a first-onset mood-, anxiety-, or substance abuse disorder after cancer may face increased risk of cancer-specific death. In an effort to optimize quality of life and survival prospects of newly diagnosed cancer patients, our findings motivate further study of underlying mechanisms as well as closer monitoring, and treatment, of severe mental disorders among these patients.

      Funding

      This study was supported by the Swedish Cancer Society (CAN 2014/417), the Swedish Research Council for Health, Working Life and Welfare (2012-0498), the China Scholarship Council (201309370015), the Swedish Society for Medical Research (SSMF, no grant number is applicable), Karolinska Institutet Distinguished Professor Award (2368/10-221 to HOA), and the Karolinska Institutet Senior Researcher Grant and the Strategic Research Area in Epidemiology (to FF, no grant number is applicable).

      Disclosure

      The authors have declared no conflicts of interest.
      Swedish Cancer Society CAN 2014/417
      Swedish Research Council for Health, Working Life and Welfare 2012-0498
      China Scholarship Council 201309370015

      References

        • Mitchell A.J.
        • Chan M.
        • Bhatti H.
        • et al.
        Prevalence of depression, anxiety, and adjustment disorder in oncological, haematological, and palliative-care settings: a meta-analysis of 94 interview-based studies.
        Lancet Oncol. 2011; 12: 160-174
        • Vin-Raviv N.
        • Hillyer G.C.
        • Hershman D.L.
        • et al.
        Racial disparities in posttraumatic stress after diagnosis of localized breast cancer: the BQUAL study.
        J Natl Cancer Inst. 2013; 105: 563-572
        • Fang F.
        • Fall K.
        • Mittleman M.A.
        • et al.
        Suicide and cardiovascular death after a cancer diagnosis.
        N Engl J Med. 2012; 366: 1310-1318
        • Chida Y.
        • Hamer M.
        • Wardle J.
        • Steptoe A.
        Do stress-related psychosocial factors contribute to cancer incidence and survival?.
        Nat Clin Prac Oncol. 2008; 5: 466-475
        • Prasad S.M.
        • Eggener S.E.
        • Lipsitz S.R.
        • et al.
        Effect of depression on diagnosis, treatment, and mortality of men with clinically localized prostate cancer.
        J Clin Oncol. 2014; 32: 2471-2478
        • Watson M.
        • Haviland J.S.
        • Greer S.
        • et al.
        Influence of psychological response on survival in breast cancer: a population-based cohort study.
        Lancet. 1999; 354: 1331-1336
        • Wikman A.
        • Ljung R.
        • Johar A.
        • et al.
        Psychiatric morbidity and survival after surgery for esophageal cancer: a population-based cohort study.
        J Clin Oncol. 2015; 33: 448-454
        • Batty G.D.
        • Russ T.C.
        • Stamatakis E.
        • Kivimaki M.
        Psychological distress in relation to site specific cancer mortality: pooling of unpublished data from 16 prospective cohort studies.
        Br Med J. 2017; 356: j108.
        • Epplein M.
        • Zheng Y.
        • Zheng W.
        • et al.
        Quality of life after breast cancer diagnosis and survival.
        J Clin Oncol. 2011; 29: 406-412
        • Lu D.
        • Andersson T.M.
        • Fall K.
        • et al.
        Clinical diagnosis of mental disorders immediately before and after cancer diagnosis: a Nationwide Matched Cohort Study in Sweden.
        JAMA Oncol. 2016; 2: 1188-1196
        • Barlow L.
        • Westergren K.
        • Holmberg L.
        • Talback M.
        The completeness of the Swedish Cancer Register: a sample survey for year 1998.
        Acta Oncol. 2009; 48: 27-33
        • Ludvigsson J.F.
        • Andersson E.
        • Ekbom A.
        • et al.
        External review and validation of the Swedish national inpatient register.
        BMC Public Health. 2011; 11: 450.
        • Klengel T.
        • Binder EB.
        Epigenetics of Stress-related psychiatric disorders and gene x environment interactions.
        Neuron. 2015; 86: 1343-1357
        • Berrino B.
        In Condensed TNM for Coding the Extent of Disease..
        Recommendations issued by ENCR: The European Network of Cancer Registries (ENCR). 2002;
        • Pirl W.F.
        • Greer J.A.
        • Traeger L.
        • et al.
        Depression and survival in metastatic non-small-cell lung cancer: effects of early palliative care.
        J Clin Oncol. 2012; 30: 1310-1315
        • Levav I.
        • Kohn R.
        • Iscovich J.
        • et al.
        Cancer incidence and survival following bereavement.
        Am J Public Health. 2000; 90: 1601-1607
        • Ranjan A.
        • Gupta P.
        • Srivastava SK.
        Penfluridol: an antipsychotic agent suppresses metastatic tumor growth in triple-negative breast cancer by inhibiting integrin signaling axis.
        Cancer Res. 2016; 76: 877-890
        • Chubak J.
        • Buist D.S.
        • Boudreau D.M.
        • et al.
        Breast cancer recurrence risk in relation to antidepressant use after diagnosis.
        Breast Cancer Res Treat. 2008; 112: 123-132
        • Reiche E.M.
        • Nunes S.O.
        • Morimoto HK.
        Stress, depression, the immune system, and cancer.
        Lancet Oncol. 2004; 5: 617-625
        • Filipski E.
        • King V.M.
        • Li X.
        • et al.
        Host circadian clock as a control point in tumor progression.
        J Natl Cancer Inst. 2002; 94: 690-697
        • Sephton S.E.
        • Lush E.
        • Dedert E.A.
        • et al.
        Diurnal cortisol rhythm as a predictor of lung cancer survival.
        Brain Behav Immun. 2013; 30: S163-S170
        • Sephton S.E.
        • Sapolsky R.M.
        • Kraemer H.C.
        • Spiegel D.
        Diurnal cortisol rhythm as a predictor of breast cancer survival.
        J Natl Cancer Inst. 2000; 92: 994-1000