SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix

Aim:  The objective of the study was to examine the impact of WWII-related caloric restriction (CR) on subsequent breast cancer (BC) risk based on individual exposure experiences and whether this effect was modified by age at exposure.

Methodology:  We compared 65 breast cancer patients diagnosed between 2005–2010 to 200 controls without breast cancer who were all members of various organizations for Jewish WWII survivors in Israel. All participants were Jewish women born in Europe prior to 1945 who lived at least 6 months under Nazi rule during WWII and immigrated to Israel after the war. We estimated CR using a combined index for hunger and used logistic regression models to estimate the association between CR and BC, adjusting for potential confounders.

Results:  Women who were severely exposed to hunger had an increased risk of BC (OR=5.0, 95% CI= 2.3–10.8) compared to women who were mildly exposed. The association between CR and BC risk was stronger for women who were exposed at a younger age (0–7 years) compared to the risk of BC in women exposed at ≥ 14 years (OR= 2.8, 95% CI=1.3–6.3).

Conclusions:  Severe exposure to CR is associated with a higher risk for BC decades later, and may be generalized to other cases of severe starvation during childhood that may have long-term effects on cancer in adulthood.


What’s known

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix
  •  The relationship between caloric restriction and cancer risk is a complex one.
  •  Most of the studies in humans were based on hunger exposure during extreme wartime situations, not controlled caloric restriction.
  •  Caloric exposure was assed based on geographical variables relating to the whereabouts of individuals during the time of exposure.
  •  In a previous Israeli study, the incidence of all cancers was higher among Israeli Jews who were potentially exposed to the Holocaust than among those who were not.

What’s new

  •  We assessed individual exposure to caloric restriction based on three indices that were constructed. Though based on previously used research tools, they added novel dimensions in order to achieve a more comprehensive view of the exposures experienced by the survivors of WWII throughout the total war period.
  •  Women who were severely exposed to hunger had an increased risk of Breast cancer.
  •  The association was stronger for women who were exposed at a younger age.

Introduction

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix

Most studies regarding caloric restriction (CR) and breast cancer (BC) risk in humans were based on war-related extreme situations, and showed inconsistent findings. The findings of four Norwegian studies (1–4) indicated that CR and malnutrition reduced the risk of BC. Concurrently, Dirx et al. detected no association between CR and BC in Dutch women exposed to the “Dutch Hunger Winter” in 1944–1945 (5). However, in another study in a different cohort of Dutch women who were exposed to the Dutch famine in the winter of 1944–1945, Elias et al. showed that a short period of intense caloric restriction at a young age led to an elevated risk of breast cancer later in life. This effect was most prominent in women who had been exposed before breast development, at an age of 2–9 years old (6). Three other studies found that World War II (WWII) related CR was associated with an increased risk of BC decades later (7–9). Most of these studies used ecological exposure data, i.e., used the whereabouts of individuals during WWII to estimate CR exposure status and did not account for potential changes in the exposure during the relevant period (1,2,5,7–9).

In a previous study, we used ecological exposure data (9) to assess overall cancer incidence among a cohort of European-born Jewish Israeli residents. We compared the incidence of cancer in a group who were potentially exposed to the Holocaust with that in the non-exposed group to determine whether the exposed group had a higher than expected all-site cancer incidence. Standardised incidence ratios (SIRs) ranged between 1.13 and 2.40 in different birth cohorts. Additionally, we found a clear gradient of risk by age at exposure, where the highest risk estimates were observed for the youngest birth cohort (SIR = 2.4, 95% CI: 2.24, 2.56). Breast cancer risk showed a similar pattern (9).

To further investigate these findings and control for potential confounders and personal risk factors, we conducted a case-control study among Jewish WWII survivors residing in Israel. In the present study, we examined whether WWII-related CR affected subsequent BC risk based on individual exposure experiences and whether this effect was modified by age at exposure.

Subjects and methods

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix

Study population

The study population comprised Israeli women who were exposed to the Nazi regime during WWII. Inclusion criteria for the study were being a current Israeli Jewish resident born in Europe prior to 1945 and having lived under the Nazi regime during WWII (years 1939–1945) for at least 6 months. Exclusion criteria included a previous diagnosis of primary cancer (excluding squamous cell and basal cell skin carcinoma), dementia or Alzheimer’s disease, and immigration to Israel after 1989. In that year and the few years that followed, Israel experienced a massive wave of immigration from the former Soviet Union (FSU). As post-war exposures of the newer immigrants may have been very different from the post-war exposures of those who immigrated to Israel between 1945 and 1989, and since information regarding these potential exposures was mostly lacking, these women were excluded. According to the Israeli Central Bureau of Statistics (ICBS), 85% of the target study population immigrated to Israel before 1960 and only 15% immigrated to Israel between 1960 and 1989 (10). Study participants were recruited between August 20th, 2007 and June 1th, 2010.

Cases were breast cancer patients diagnosed between 2005 and 2010, with a histological confirmation (based on a review of their medical records) of in-situ or invasive breast malignant tumour. The cases were recruited through the oncology and radiology departments, as well as the out-patient oncology clinics, in five medical centres in Israel (Rambam Health Care campus, Haifa; Sourasky Medical Center, Tel-Aviv; Rivka Ziv Medical Center, Safed; Western Galilee Hospital, Nahariya; Laniado Medical Center, Natanya). Based on identified medical records from the medical centres and the Israeli Cancer Registry, 111 women met the inclusion criteria of the study and 65 of them (58.6%) agreed to participate. The mean time between diagnosis and the interview was 2.36 years (SD = 1.42).

The controls were located through various voluntary assistance organisations for Jewish WWII survivors, such as “Amcha”, the Kibbutz Movement (since in Israel many WWII survivors live in kibbutzim), local welfare departments, retirement homes and special hostels for WWII survivors and the foundation for the benefit of the holocaust victims in Israel. Among the control group, 368 women met the inclusion criteria and 204 (55.4%) agreed to participate. Four interviews were excluded due to inability of the interviewees to remember personal historical data. No marked differences were observed between the group of women who agreed to participate in the study and the group of women who refused, with regard to age, current residence and country of birth in Europe.

In total, 65 BC patients and 200 controls took part in the study. The study was approved and supervised by the Ethical Review Board in each of the medical centres that participated

Study instruments

Following a short explanation regarding the study and its aims, each of the participants was asked to sign an informed consent form. All cases and controls were interviewed in their homes by a trained interviewer. The questionnaire collected self-reported demographic data (i.e., age, country of birth, marital and employment status, education, income, degree of religious observance), health behaviour data (i.e., current BMI, physical activity, smoking status, alcohol consumption), obstetrical and gynaecological factors (i.e., age at menarche, menstrual patterns, number of pregnancies, number of live-born children, age at first child birth, age at last child birth, difficulty conceiving, using fertility treatment, overall duration of breastfeeding, use of oral contraceptives and menopausal characteristics such as age at menopause, hormonal replacement therapy, surgical operation and surgical menopause, family history of breast and ovarian cancer including BRCA1 and BRCA2 mutations). Additionally, the questionnaire referred to data that enabled the ranking of the exposure status of the participants with respect to caloric restriction during WWII, as explained herewith.

Exposure ascertainment

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix

A number of indices were designed to rank the exposure to CR during WWII, and were described elsewhere (11). Briefly, three indices were developed:

  •  Hunger exposure score: The score was calculated for each participant based on her reported whereabouts during WWII years (e.g. ghettos, work/concentration/death camps, living under false identity, hiding away/shelter or living in the open (forests, wandering) etc) which were ranked by the level of CR experienced in each of them, based on historical data and on the pilot study results. These ranks were subsequently multiplied by the amount of time spent in each location (in months), covering the total period of WWII (1 September 1939–8 May 1945), and then summed up across locations, yielding an individual, continuous hunger exposure score that took into account the secular changes in the exposure throughout the War period (11). Additionally the hunger score was categorised to “mild hunger,”“moderate hunger,” or “severe hunger”, based on tertiles (Appendix A presents an example of the calculation of the hunger score index).
  •  Hunger symptoms score: Each participant was asked whether she had experienced any of 17 hunger-related symptoms and signs in each of the different staying locations during WWII. The following symptoms and signs were considered hunger-related: weight loss, watery diarrhoea, abdominal oedema, oedema of the feet and the hands, polyuria, vitamin deficiencies (such as scurvy, rickets, night blindness), anaemia, goitre, amenorrhea or irregular menses, hirsutism and voice change (12). Each symptom/sign reported was counted only once, and all symptoms/signs were summed to form a hunger symptoms score (ranging from 0 to 17) (11).
  •  Self- perceived hunger score: The participant was asked whether she experienced hunger in each of the different staying locations during WWII and was asked to rank it. The scale ranged between 1 = not at all, and 4 = to a great extent. The score was averaged over all WWII whereabouts to form the self-perceived hunger score (6,11).
  • However, since only the first (‘Hunger exposure score’) of these indices proved to be sensitive enough to detect differences in the exposure to hunger among the participants, we will report only on this index in the results.

Other related data

The medical files of the breast cancer patients were abstracted in order to investigate their disease characteristics.

Statistical analysis

Frequencies of categorical variables (or means and standard deviation) in the case of continuous variables such as age were computed separately for cases and controls. The frequencies were cross-tabulated and differences between cases and controls were statistically assessed using the χ2 test. Continuous variables were statistically assessed by using independent t-tests. In the case group, we examined the relationship between age at the time of diagnosis, hormone receptors, HER2 status, stage at diagnosis, BRCA1 and BRCA2, and hunger exposure.

Differences between the case and the control subgroups were tested by χ2 analyses for categorical variables, and by one-way ANOVA tests for continuous variables. For variables showing overall significant differences, additional comparisons between groups were made using Scheffe’s test for post hoc contrasts.

We examined the relationship between each WWII CR index (e.g., hunger exposure score, hunger symptoms score and self-perceived hunger score) separately and BC risk. Additionally, we stratified for age during WWII.

Multivariate logistic regression models where the dependent variable was breast cancer (yes or no) were constructed in two steps. First, the independent variables that were found to be statistically significant in predicting the dependent variable in univariate analyses were added to the model separately: the hunger exposure score, age at the beginning of WWII, education attained, BMI. Then, we carried out multiple logistic regression models that included the individual hunger score, age at the time of WWII, and variables that were statistically significant in the former analyses, such as education and BMI, However, BMI was not found to be significant and therefore was removed from the final model. In all models, odds ratios (ORs) and 95% confidence intervals (CIs) were calculated and statistical significance was set as p ≤ 0.05.

Additionally, inter-relationships were tested between age at the beginning of WWII as an interaction variable in the association between caloric restriction and BC. All analyses were performed with SPSS 18.0.0 for Windows 7 64bit (SPSS Inc., Chicago, IL, USA).

Results

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix

Baseline characteristics of the study groups are summarised in Table 1. The mean age at interview in the BC cases (76.2 ± 5.6) was lower compared to the controls (78.3 ± 5.6). In the case group most women reported having academic degrees (32.3%) while in to the control group most women reported having no degree at all (46.5%). No marked differences were observed between the cases and controls with respect to monthly household income, level of religious observance and marital status.

Table 1.   Baseline characteristics and WWII experiences for breast cancer cases and controls among Jewish WWII survivors living in Israel, 2007–2010
CharacteristicCases (n = 65)Controls (n = 200) p-Value
  1. Note: NIS is equivalent to 1173 USD.

  2. *The income rank was based on the average gross monthly income per + 65 year old subjects in Israel which is 4439 NIS (equivalent to 1300 USD). †Country of birth was categorised by its status during WWII: Under Direct Nazi Regime: Germany, Austria, Czechoslovakia, Hungary, Poland, Lithuania, Latvia, Belgium, Holland France and Yugoslavia. “Independent countries” = countries that cooperated with the Nazi regime: Romania, Bulgaria, and Italy. Tertiles.

Socio-demographic
Age at interview (Mean ± SD)76.2 ± 5.678.3 ± 5.6< 0.001
  N (%) N (%)  
Education
 No degree17 (26.2)93 (46.5)< 0.001
 High school degree27 (41.5)79 (39.5) 
 Bachelor, masters or doctoral degree21 (32.3)28 (14.0) 
Marital status
 Married/cohabitating62 (95.4)198 (99.0)0.097
 Single/divorced/widowed3 (3.4)2 (1.0) 
Religion observance level
 Secular61 (93.8)170 (85.0)0.086
 Traditional/religious/orthodox4 (6.2)30 (15.0) 
Income*
 < 4000 NIS7 (10.8)38 (19.0)0.291
 ≥ 4000 NIS47 (72.3)128 (64.0) 
 Refused to answer11 (16.9)34 (17.0) 
Health related behaviours
Alcohol consumption
 Ever9 (13.8)49 (24.5)0.071
 Never56 (86.2)151 (75.5) 
Cigarette Smoking status
 Ever28 (43.1)80 (40.0)0.661
 Never37 (56.9)120 (60.0) 
Current physical activity (≥ 20 m/week)36 (55.4)137 (68.5)0.054
Body mass index (kg/m2)
 BMI ≤ 2521 (35.6)65 (36.1)0.014
 Over weight (25 ≤ BMI < 30)18 (30.5)84 (46.7)  
 Obese (BMI ≥ 30)20 (33.9)31 (17.2)  
WWII experiences
Country of birth†
 Under direct Nazi regime41 (63.1)167 (83.5)< 0.001
 “Independent country”24 (36.9)33 (16.5)  
Age at the time of WWII (years)
 0–732 (49.2)54 (27.0)< 0.001
 8–1320 (30.8)54 (27.0) 
 ≥ 1413 (20.2)92 (46.0) 
Born during WWII8 (12.3)15 (7.5)0.308
Hunger exposure‡
 Mild hunger14 (21.5)50 (25.0)< 0.001
 Moderate hunger11 (16.9)76 (38.0)  
 Severe hunger40 (61.5)74 (37.0)  
Hunger symptoms score (Mean ± SD)1.9 ± 2.21.8 ± 1.80.715
Self-perceived hunger score (Mean ± SD)2.8 ± 1.22.5 ± 1.10.241
Hunger exposure score (Mean ± SD)167.3 ± 69.8119.5 ± 53.2< 0.001

Most cases and controls reported that they were under direct Nazi rule (63.1% and 83.5%, respectively, p < 0.001). Eight (12.3%) women were born during WWII in the case group compared to 15 (7.5%) in the control group, (p = 0.308). Regarding health-related behaviour characteristics, women in the control group were more likely to be overweight, while women in the case group were more likely to be obese (p = 0.014), and no significant differences were observed with respect to smoking status, alcohol consumption and current physical activity (Table 1). No significant differences were observed between the cases and controls regarding reproductive and gynaecological characteristics (Table 2).

Table 2.   Reproductive and gynaecological characteristics for breast cancer cases and controls among Jewish WWII survivors living in Israel, 2007–2010
CharacteristicCasescontrolsp-Value
[N = 65][N = 200]
  1. *Mean difference between first child birth to last child birth.

Age at Menarche (years) [mean ± SD]13.45 ± 1.7313.79 ± 1.880.200
Regular Menstrual pattern [n (%)]51 (78.5)167 (83.5)0.355
Ever pregnant [n (%)]61 (93.8)194 (97)0.246
Number of pregnancies [mean ± SD]3.96 ± 2.123.41 ± 1.700.060
Number of live-born children [mean ± SD]2.44 ± 0.952.64 ± 1.050.217
Difficulty conceiving [n (%)]8 (12.3)47 (23.5)0.053
Age at first childbirth (years) [mean ± SD]24.21 ± 3.9923.51 ± 4.000.250
Age at last childbirth (years) [mean ± SD]31.47 ± 4.0831.29 ± 4.630.801
Births’ gap* (years) [mean ± SD]7.7 ± 4.018.04 ± 4.350.629
Average of total breastfeeding (months) [mean ± SD]10.45 ± 10.319.91 ± 10.570.738
Age at Menopause (years) [mean ± SD]49.95 ± 5.7248.69 ± 5.010.117
Ever underwent gynaecological surgery [n (%)]26 (40)66 (33)0.368
Surgical menopause [n (%)]12 (46.2)45 (68.2)0.060
Ever use of sex hormones
Use of oral contraceptive pill [n (%)]7 (10.8)22 (11)0.959
Use of fertility treatment [n (%)]2 (3.1)17 (8.5)0.174
Use of hormonal replacement therapy [n (%)]24 (36.9)50 (25)0.080

With regard to WWII experiences, cases reported a younger age at the beginning of WWII (p = 0.003) and were more likely to report being under direct Nazi rule as opposed to living in a country that collaborated with Nazi Germany, compared to the controls (p < 0.001) (Table 1). Furthermore, the breast cancer cases had a higher rank of hunger exposure score (i.e., were more exposed to CR) compared to controls (mean of 167.33 vs. 119.54, p < 0.001), and were frequently categorised as being under severe hunger (61.5%) during the war, while the controls were more likely to be categorised as being under moderate hunger (38%). Tumour characteristics were not significantly associated with the CR exposure experienced, based on the analyses done in the case group (Table 3).

Table 3.   Association between age at the time of diagnosis, hormone receptors, HER2 status, stage, and hunger exposure for the breast case group in Jewish WWII survivors living in Israel, 2007–2010
 Mild Hunger (n = 14)Moderate hunger (n = 11)Severe hunger (n = 40)p-value
  1. Note: *For variables showing overall significant differences, additional comparisons between groups were made using Scheffe’s test for post hoc contrasts; significant differences was found between the group of Moderate hunger to Mild hunger p < 0.029.

  2. Range 64–88 years.

Age at the time of diagnosis (years)† (Mean ± SD)71.14 ± 5.5877.45 ± 3.9074.05 ± 6.13 0.029*
  N (%) N (%) N (%)  
Stage (n = 63)
 02 (14.3)2 (18.2)2 (5.3)0.534
 15 (35.7)4 (36.4)15 (39.5) 
 24 (28.6)0 ( 0.0)10 (26.3) 
 32 (14.3)3 (27.3)7 (18.4) 
 41 (7.1)2 (18.2)4 (10.5) 
ER status (n = 55)
 Positive11 (78.6)7 (77.8)25 (78.1)0.978
 Negative3 (21.4)2 (22.2)7 (21.9) 
PR status (n = 55)
 Positive11 (78.6)6 (66.7)23 (74.2)0.832
 Negative3 (21.4)3 (33.3)8 (25.8) 
HER2 (n = 50)
 Positive3 (25.0)2 (25.0)8 (26.7)0.904
 Negative9 (75.0)6 (75.0)22 (73.3) 
BRCA 1 (n = 31)
 Positive 
 Negative8 (100.0)4 (100.0)19 (100.0) 
BRCA 2 (n = 31)
 Positive1 (25.0)0.681
 Negative8 (100.0)3 (75.0)19 (100.0) 

Multivariate logistic regression models were carried out in two ways, as described earlier. First, age at time of WWII, education, current BMI and hunger exposure score were included in the model by a stepwise approach. The results indicated that current BMI was not a significant predictor once adjusted for the other confounders and therefore was removed from the model. Then, we constructed a final model based on age at the beginning of WWII, education, and hunger exposure, which indicated a significant impact for all predictors (Table 4).

Table 4.   Odds ratios for breast cancer according to hunger exposure, age at the time of WWII for breast cancer cases and controls in Jewish WWII survivors living in Israel, 2007–2010
FactorsOR95% CIp-valueOverall p-value
Age at the time of WWII (years)
 0–72.81.3–6.30.0120.043
 8–131.80.8–4.10.176 
 ≥ 14Reference  
Education
 High school degree1.50.7–3.20.2680.017
 Bachelor, masters or doctoral degree3.41.4–8.10.004  
 No degreeReference  
Hunger exposure score
 Moderate hunger1.20.5–2.90.752< 0.001
 Severe hunger5.02.3–10.8< 0.001  
 Mild hungerReference  

The risk of breast cancer increased with increasing hunger exposure. Women who were severely exposed to the hunger had a statistically significantly higher risk of breast cancer (OR = 5.0, 95% CI = 2.3–10.8) compared to women who were mildly exposed.

The largest impact of age at the beginning of WWII on breast cancer risk was seen for women who were exposed at a young age (0–7 years) compared with the risk of breast cancer in women who were exposed at ≥ 14 years (OR = 2.8, 95% CI = 1.3–6.3), the risk in women who were exposed to WWII between the ages of 8–13 years was not statistically significant OR = 1.8 (95% CI = 0.8–4.1).

We conducted separate analysis, excluding the six in-situ cases. The results were essentially the same (data are not shown).

The inter-relationships between age during the beginning of the WWII and CR and the extent to which they are independently associated with BC risk were assessed in separate analyses (using stratification techniques or multivariable logistic models as appropriate); no statistically significant interactions were detected.

Discussion

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix

The aim of the current study was to assess the impact of childhood CR on later BC risk. Its main results indicate that WWII-related CR, as measured by the combined hunger score, is associated with an increased risk for subsequent BC. Age at exposure to CR during WWII was inversely associated with BC risk, i.e., the younger the age at WWII, the higher the risk.

Our findings are not totally in line with animal experiments, where CR is consistently associated with cancer prevention (13–18). One reason for that may be the basic differences between experimental CR in laboratory animals and WWII related CR among European Jews; during the War time, the survivors suffered ∼ 80% reduction in their usual energy intake with significant changes in diet quality (11,19–21), while experimental CR in laboratory animals involve moderate (10%–30%) or severe (40%–80%) reduction in calories in a closely controlled environment and through a balanced diet in respect to macro- and micronutrients, and therefore CR is often referred to as “under-nutrition without malnutrition”(13). More notably, WWII survivors were exposed to CR at differing levels and durations – up to 5 years, while experimental CR is usually stable and life-long. Furthermore, WWII survivors were not only exposed to CR, but to many other factors concurrently that could have interacted with CR to produce changes that eventually affected their individual susceptibility to cancer.

Our findings are in accordance with several former human-based WWII related CR studies (6–8,22), but in contrast with others (1,2,4). The different results may be partly explained by the different methodologies, but may also reflect a modification of the outcome by other variables, such as severity of the exposure to CR or its length. In Norway, for example, there was an overall reduction in the average daily caloric intake, from 3475 kcal in 1939, to a minimum of 2700 kcal in late 1944 and early 1945 (∼ 22% reduction in calorie intake without significant changes in diet quality) (24). “In Guernsey, one of the British Channel islands occupied by the Nazis throughout WWII, the diet of the islanders during the first 3 years of the Occupation (1940–1943) was similar to that in the UK at the time; that is, food was available but rationed. However, after the Allied invasion of France in June 1944, the Island’s access to food supplies was severely limited and the population lived on much reduced rations, estimated at approximately 1200 calories per day per person. This state of semi-malnutrition persisted until the island was liberated in May 1945, although some relief (about extra 460 calories per person per day) was brought through Red Cross parcels from December 1944 onwards” (7). In the Netherlands, the average daily intake was around 800 Kcal/day at the peak of the ‘hunger winter’ (September 1944–May 1945), but the rations were usually adequately balanced in terms of macronutrients (protein, carbohydrates and fat) (5,6) (∼ 70% reduction in rations for adults; 50% reduction in rations for children without significant changes in diet quality) (21). The Russian survivors of the siege of Leningrad were exposed to even longer term and more severe energy restriction (8,23); around 80% reduction in the usual energy intake with significant changes in diet quality (21). Similar caloric restriction was found among the European Jews during WWII as documented in the literature and in our pilot study (11,19,20). However, the main difference is the duration of the exposure to CR: the siege on Leningrad lasted 872 days, but the most severe hunger lasted 1 year (1941–1942) (23). The duration of the exposure to famine experienced by European Jews during WWII was usually longer and lasted up to 5 years.

Elias et al. (6) hypothesised that the higher susceptibility to BC observed in women severely exposed to the Dutch Famine may have been related to a permanently and irreversible impact on the endocrine system, delivered by the abrupt halt of the famine and the subsequent availability of unlimited food supply. These endocrine changes may have been modifying the risk for BC in the long run (6). The finding that circulating levels of insulin like growth factor-I (IGF-I), a hormone involved with epithelial cells turnover and associated with higher risk for postmenopausal BC (24), were statistically significantly elevated among those with the greatest exposure to the Dutch famine (25), further supported this hypothesis. While in Guernsey the return to normal nutrition was gradual (7), CR in the Russian survivors of the siege of Leningrad continued even after the end of the War (23,26). In the current study, we had no data on the time of returning to normal nutrition following the end of WWII, which probably varied from one woman to another, due to the different nature of their post-war experiences. This fact may have added an additional aspect to the complex exposure to CR that was created during WWII.

Another possible link between exposure to CR and breast cancer is through epigenetic mechanisms such as global DNA hypomethylation that may cause chromosomal instability and oncogene activation, and results in increased breast cancer risk (27).

Prenatal famine in humans has been associated with various later-life consequences, depending on the gestational timing of the insult and the sex of the exposed individual (28–32). Epigenetic mechanisms have been proposed to underlie these associations. One unique cohort referring to these associations is the Dutch Hunger Winter Families Study (30). The first candidate locus investigated in this cohort was the well-characterised differentially methylated region of the imprinted insulin-like growth factor 2 (IGF2) gene, a key factor in human growth and development (33–35). Ito et al. (33) assessed whether IGF2 DMR0 methylation is either present constitutively prior to cancer or whether it is acquired tissue-specifically after the onset of cancer in the EPIC samples. The findings disclosed that IGF2 DMR0 methylation levels in tumour cells were lower than the levels in matched non-tumour tissue samples. Hypomethylation of DMR0 was detected in breast tumour tissues (33%).

Individuals who were prenatally exposed to famine during the Dutch Hunger Winter in 1944–1945 had, 6 decades later, less DNA methylation of the imprinted IGF2 gene compared with their unexposed, same-sex siblings (36). Animal studies and their early human data on the imprinted IGF2 locus indicated a link between prenatal nutritional and DNA methylation (36). However, it remains unclear how common changes in DNA methylation are and whether they are sex- and timing-specific paralleling the later-life consequences of prenatal famine exposure (32).

The role of age at the time of CR in BC aetiology is of interest and further implies a complex mechanism. The Dutch researchers (6) reported that the association between famine exposure and BC risk was stronger in women who were exposed between the ages of 2 and 9 years than for women who were exposed at older ages. Fentiman et al. (7) revealed a non-significantly higher incidence in the subgroup born from 1926 to 1934 and aged 10–18 in 1944, at the peak of the occupation-related CR (13). This age range is also consistent with the results of the Leningrad Siege studies, which indicated a significantly higher BC mortality in the group of women who were exposed to CR when they were 10–18 years of age (8).

Age at exposure modified the outcome in previous studies (3,6–9) but not in the current study. As we summarised in our previous study (see Ref. (9). “Early exposures, including antenatal ones, have previously been proposed (37–42) as modifiers of the individual susceptibility for future chronic morbidity. The mechanisms involved possibly include long-term impact on growth patterns, sensitivity of hormone receptors, basic hormonal levels, and behavioural responses that might alter long term susceptibility to certain diseases (37–40). Age at exposure is a known modifier of cancer incidence, as shown by results of follow-up studies (41,42) in which associations have been found between childhood exposure to therapeutic radiotherapy for tinea capitis, enlarged tonsils, or thymus gland and an increased risk of cancers of the thyroid, salivary gland, central nervous system, skin, and breast, as well as leukaemia. Likewise, an excess risk of breast cancer has been reported among scoliosis patients who had frequent diagnostic x-rays during childhood and adolescence (41). Similar findings were reported from a cohort of atomic bomb survivors, among whom younger age at exposure was inversely associated with risk of a solid malignant tumour (42)”. The women in the case group in the current study were, on average, 7.9 years old at the beginning of WWII (range: 0–20), significantly younger than the women in the control group (average age: 11.4, range: 0–28).

This study has limitations, and its results should therefore be interpreted cautiously. Since most of the data collected in this study was self-reported and referred to events that took place 65 years ago, the results may be biased by recall. Still, the exposure data was based not on what was actually consumed but on the whereabouts of each participant, which are probably easier to remember. Additionally, the testimonies were in accordance with historical data and published literature.

Of the three indices used to assess CR exposure, full use was made only of the individual hunger score, since the limited range of the other two indices created a variance that was too small to detect significant differences between the cases and the controls. However, the results of our pilot study (17) showed that the individual hunger score was closely and significantly correlated with the hunger symptoms score (11).

The study has some important strengths as well. Focusing on WWII- exposed women only did limit our ability to detect differences in CR exposure between the cases and the controls but helped to avoid potential selection bias. Furthermore, the hunger exposure score, which was the main exposure measurement tool, was based on whereabouts during WWII, not dietary intake, which are easier to remember and less prone to biases, and took into account the changes in the exposure throughout the war years. This was never done before, as all exposure tools used in similar studies referred to a single, dichotomous variable and were not able to capture the cumulative nature of this CR exposure.

In conclusion, the results of the current study support our former results as well as the results of other studies based on non Jewish populations exposed to WWII-related CR (6–8) and indicate an independent effect (9) of exposure to severe CR on subsequent breast cancer, diagnosed many years later. Age at exposure plays a role in this association. This finding is of relevance not only to WWII survivors, but to other communities that may be currently exposed to severe CR. It is also of importance when cancer aetiology is considered, and the biologic mechanisms behind it warrant further confirmation.

Acknowledgements

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix

The authors are indebted to the participating women for volunteering to take part in this study. They would also like to express their gratitude to Prof. S. Shasha for his contribution to this field of research in general and to this study in particular.

The current manuscript is based on the Ph.D. thesis of Neomi Vin-Raviv, which was submitted to the School of Public Health, the University of Haifa, Haifa, Israel, in partial fulfilment of the requirements for the Ph.D. Degree.

Author Contributions

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix

N. Vin-Raviv wrote the manuscript, she carried out the work described in this article and the main statistical analyses described in this study.

L. Keinan-Boker assisted N. Vin-Raviv with the analysis of the data and the writing of this article.

L. Keinan-Boker, M. Barchana and S. Linn were involved with formulation of the main study idea and hypotheses and supervised N. Vin-Raviv Ph.D. Dissertation on which this article is based.

References

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix

Appendix

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Subjects and methods
  6. Exposure ascertainment
  7. Results
  8. Discussion
  9. Acknowledgements
  10. Author Contributions
  11. References
  12. Appendix

Appendix A: an example of the calculation of the hunger exposure score

The calculation of the hunger exposure score was based on the following principles: First, an exact time line of the different staying locations during WWII (years 1939–1945) was built per each participant, based on self reporting. The time (in months) spent in each location was listed. Next, locations were categorised into six main groups: (i) ghettos; (ii) work/concentration/death camps; (iii) Refugees in eastern USSR; (iv) hiding away/shelter or living in the open (forests, wandering etc) under Nazi role; (v) living under false identity; (vi) movement restriction curfew areas. Each category of staying location was ranked to describe the average daily caloric intake related to it, based on previous publications (12,19,20) and on a prior qualitative study that consisted of open interviews with 10 WWII survivors. Based on these depth interviews, a semi-open basic food consumption questionnaire was constructed and then implemented in the pilot study. The 43 participants’ replies were evaluated by a trained nutritionist and compared to the minimal recommended caloric intake of 1200 Kcal/day. The participants reported the following daily calorie intake: (i) Ghettos/other camps/Refugees in eastern USSR – 200–300 Kcal/day, a quarter of the minimal daily recommended caloric allowance (0.25). This daily food intake was imbalanced in terms of macronutrients, and included mostly carbohydrates and fats and hardly any proteins. (ii) Hiding/staying with resistant fighters/wandering – 500–600 Kcal/day, half of the minimal daily recommended caloric allowance (0.5). The daily food intake consisted on seasonal fruit or vegetables which were available in the forests or rural areas. (iii) Living under false identity/movement restriction/curfew – 1200 Kcal/day, full minimal daily recommended caloric allowance (1.0). The diet was similar to that of the non-Jewish population. Thus, the following ranks have been allocated, using the reciprocal values of the numeric ones, in order to adjust the scale: (i) Ghettos/camps/Refugees in eastern Soviet Union = score 4 (severe hunger); (ii) Hiding/staying with partisans/wandering = score 2.5 (moderate hunger); (iii) living under false identity/ movement restriction curfew = score 1 (mild hunger). The ranks were subsequently multiplied by the amount of time spent in each location, covering the total period of WWII (1.9.1939–8.5.1945), and then summed up across locations, yielding the individual hunger exposure score.

  Participant 3Participant 7Participant 139Participant 263
  1. *All calculations are made by month.

 1WWII official start date1/9/19391/9/19391/9/19391/9/1939
 2WWII individual start date15/4/19411/19415/10/19401/3/1939
 3Gap* between official and individual start date (2–1)191880
 4Rank1111
 5Score (3 × 4)191880
 6WWII individual liberation date15/4/19451/9/19448/5/19451/1/1945
 7WWII official end date8/5/19458/5/19458/5/19458/5/1945
 8Gap* between official and individual end date (7–6)1804
 9Rank1111
10Score (8 × 9)1804
11Total time* at ghettos   28
12Rank   4
13Score (11 × 12)112
14Total time* at camps30  36
15Rank4444
16Score (14 × 15)120144
17Total time* as refugees in eastern soviet union    
18Rank4444
19Score (17 × 18)
20Total time* in hiding away/shelter or living in the open under Nazi role  60 
21Rank2.52.52.52.5
22Score (20 × 21)150
23Total time* at movement restriction curfew areas 42  
24Rank1111
25Score (23 × 24)42
26Total time* at false identity180 18
27Rank1111
28Score (26 × 27)18018
29Hunger exposure score (5 + 10 + 13 + 16 + 19 + 22 + 25 + 28) 158 68 158 260