Breast cancer in Iraq is associated with a unimodally distributed predominance of luminal type B over luminal type A surrogates from young to old age

Background Breast cancer has recently increased in post-menopausal Iraqi women. In Western countries at high-risk for breast cancer, there is a bimodal increase in estrogen receptor (ER) positive tumors with a peak of low proliferation rate luminal A over higher proliferation rate luminal B tumors after 60 years of age. The aim of this study was to analyze in Iraqi women whether shifts are occurring in immunohistochemical (IHC) surrogates of molecular breast cancer subtypes toward a high-risk profile. Methods Age specific and age standardized womens breast cancer incidence was estimated for the years 2006 through 2012. IHC results of ER, PR, HER2, and Ki67 testing were analyzed on the breast cancers of 125 Arabic and 725 Kurdish women by frequency of distribution and by age. Results Between 2006 and 2012, age standardized incidence of breast cancer in Iraq increased from 30 to 40/100,000 women with the increase specifically occurring in women ≥ 60 years old (P < 0.001). Breast cancers in Kurdish women ≥ 60 years old may also have increased (P = 0.047) with urban exceeding rural rates by 2:1. For both Kurdish and Arabic women, there was a marked predominance of luminal B tumors at all ages in which luminal B and luminal A tumors were asymmetric skewed toward older age but with no late luminal A age peak. Conclusions Older Iraqi women do not show the bimodal shift toward higher rates of luminal A breast cancers seen in the West. The modest increase in age standardized incidence of breast cancer in Iraqi is being seen specifically in older women and may be better attributed to a trend for care in urban cancer centers rather than changing tumor characteristics.


Background
In the developing world, cancer is projected to increase by 70% over the next 20-25 years. Breast cancer represents 20-30% of cancer among women and is likely to account for a major part of that increase [1]. These expectations are anticipated because the populations of developing countries are ageing and cancer is largely a disease of older people. Life-style changes are likely to contribute, because transitions from a traditional to a Westernized life-style expose women to higher individual risk [2].
The world can be divided into low-risk and high-risk regions on the basis of differences in breast cancer incidence [1,2]. Those at high-risk consists of the developed countries of Northern Europe, North American, Uruguay, Argentina, Australia and New Zealand. In these areas, age standardized incidence exceeds 80-100/100,000 women. In the developing world, age standardized incidence (ASI) varies from 20-60/100,000 women, and in parts of Central and Southeast Asia and sub-Saharan Africa, estimates are less than 20/100,000 [1,2].
Breast cancer in all parts of the world begins to be seen at about 20 years of age [3]. Afterward, age specific incidence steadily rises until the menopause when geographic differences begin to be seen. In the developing world, a flattening and then a decline occur about 10-15 years after the menopause. In developed countries, the age specific incidence accelerates after the menopause, and at age 70 and above doubles that seen at 45-49 years old [3][4][5]. The life-style changes contributing to increased post-menopausal breast cancer are largely premenopausal and include obesity, low rates of childbirth, infrequent or no lactation, early menarche, and late menopause [2]. These factors, that are common in highrisk countries, promote a state of relative hyperestrogenism and the development of estrogen responsive tumors [3][4][5]. The life-style factors are becoming more common in countries considered low-risk particularly in their growing urbanized communities [6].
Molecular subtyping has defined at least four categories of breast cancer with the most prevalent type and that being most closely related to increased estrogen exposure being estrogen receptor (ER) positive and negative for the Her-2-neu (HER2) proto-oncogene [3]. ER positive tumors are further subtyped into low proliferation rate luminal A and higher proliferation rate luminal B tumors with the luminal A tumors being more frequently seen in older women in regions of the world at high-risk for breast cancer [3].
Breast cancer rates in Iraq and its Kurdish region were generally stable between 2000 to 2009, but newer data from the Iraqi Cancer Registry reveal rising rates since 2009 with women after age 50 making the major contribution to the increase [5,[7][8][9]. In this study, demographic changes are explored together with tumor ER and HER2 subtyping. The aim of the study was to determine whether age related shifts may be occurring in ER and HER2 subtypes that may offer insight into potential increases of breast cancers in culturally transitional Middle-Eastern countries.

Calculation of age specific and age standardized incidence
Data for the country of Iraq were obtained from the publications of the Iraqi Cancer Registry for the years 2006 and 2012 [9,10]. These publications provided detailed information regarding the age specific incidence rates of breast cancer and the female population of Iraq in the 5-year age ranges from 0 to 70+ years old. The proportions of the population in the 5 year age ranges correspond to that published by the United Nations Population Division (population pyramid.net).
Data for Sulaimaniyah were obtained from the Hiwa Regional Oncology Center cancer registry. The data collection was limited to the years before 2014 so as to not be influenced by the large recent influx of internally displaced persons into the Kurdish region. Breast cancers for Sulaimaniyah were averaged for the three year periods 2006-2008 and 2011-2013. The 2006 and 2012 Sulaimaniyah populations were obtained from the Iraqi Cancer Registry and matched with the population estimates from the Sulaimaniyah Department of Statistics. The 5 year age distribution followed the proportions used for all of Iraq.
For each region, age specific incidence rates were calculated on the basis of the number of cancers per 100,000 women in the specified age ranges. Age standardized incidence rates were calculated using the WHO World Standard population.

Analysis of breast cancer subtypes
Immunohistochemistry (IHC) for ER, progesterone (PR), HER2, Ki67, and HER2 FISH for IHC equivocal HER2 cases was performed for breast cancers of Hiwa Hospital patients at the Shorsh Hospital Pathology Department. Details of the technical procedures have been previously published [5]. The luminal A subtype was defined as ER +/HER2-with a KI67 index ≤20% and PR+ with an Allred PR score ≥ 4; the luminal B subtype was defined as an ER+ and/or PR+ tumor, ± HER2+, and with a Ki67 index > 20% [11][12][13][14].

Statistical methods
Data was entered into an Excel worksheet and analyzed with Stata IC10™ (Stata Corp, College Station, TX) and SigmaStat 3.5 (Systat Software Inc., San Jose, CA) statistical software. Changes in breast cancer incidence in the years 2001 to 2012 were analyzed by Poisson's regression for the age ranges 20-39, 40-49, 50-59, and 60+ years old [15]. Differences between groups were tested by Kruskall-Wallace analysis of variance (ANOVA) on ranks with Dunn's post-hoc pairwise tests. Age by ethnicity was normally distributed for some receptor subtypes but not others, and two-way comparisons used Wilcoxon rank-sum tests. For all statistical procedures P < 0.05 was considered significant.

Age distribution of Iraqi and Sulaimaniyah populations
The Sulaimaniyah population estimates and incidence rates are for ethnic Kurdish women only. Table 1  Age standardized and age specific incidence Table 2 demonstrates the age specific incidence rates in the regions and the changes over the period from 2001 to 2012. Prior to 2012, the age standardized rates were approximately 30 per 100,000 women with a peak age specific incidence at 50-54 years old and a declining rate in older age groups. At 70+ years old, rates were similar to those seen at 35-39 years of age. In all of Iraq in 2012, age specific incidence exceeded 110/100,000 for all 5-year age groups from 50 to 70+ years old, and the age standardized rate increased to 40/100,000. In Sulaimaniyah, age standardized incidence increased slightly from 32 to 36/100,000 between 2006 and 2012.

Distribution of breast cancer by age
By Poissons regression, in all of Iraq over the decade 2001-2012, there was a significantly increased breast cancer rate between 2006 and 2012 among women ≥ 60 years old (P <0.001) but not in younger age groups (Table 3). Table 3 also indicates that for Sulaimaniyah, breast cancer rates may also have increased in the ≥ 60 year old age group (P = 0.047).

Rural vs urban differences
We attempted to look at urban vs rural differences in breast cancer rates. Data from Baghdad were not available. In 2011-2013, 63.1% of Sulaimaniyah breast cancer patients identified themselves as living in the central city (3-year average, central city 137 patients vs other     (Table 5). Figure 2 shows that luminal B/HER2-tumors began to be seen at an earlier age than Luminal A tumors, but both subtypes were closely nested over the entire range of young to old age with no suggestion of a late luminal A peak. Luminal B/HER2+ tumors had a significantly earlier peak than the other subtypes (P < 0.01). Although Luminal B/HER2+ tumors represented only 28% of all luminal B tumors, the earlier age contributed to all luminal B tumors being found at a significantly earlier age than luminal A tumors (P = 0.04) but with no significant difference in age between luminal A and luminal B/HER2-tumors (P = 0.19). Figure 3 plots age specific incidence per 100,000 women for the major subtypes. The peak incidence of all cancers was about 60 years of age but then declined markedly from 60 to 80 years old. The decline includes luminal A and luminal B subtypes with the incidence of luminal B exceeding that of luminal A at all ages (Fig. 4).

Discussion
Molecular breast cancer subtyping is widely used clinically to identify patients at low risk for metastasis and can be roughly inferred on the basis of IHC testing [12,13]. While the 2015 St Gallen's consensus considered the IHC designation of molecular subtypes as being impractical for guiding treatment, the assignment may be valuable from an epidemiologic standpoint [3,13]. In the West, the average age of newly diagnosed breast cancer patients is > 60 years old, and the high rates of breast cancer are the result of a several-fold increase in the age specific incidence of ER+/HER2-tumors in post-menopausal women [3][4][5].
In US women, on the basis of IHC with parallel PAM50 molecular testing, luminal B breast cancers were found to be more common than luminal A tumors by a factor of 2.48 under 40 years of age with the ratio decreasing to 1.27 at 40-49 years old [16]. At 50-59 years of age, the proportions reached parity, and at > 60 years old, luminal A exceeded luminal B tumors by 25-50% [16]. In terms of age specific density, this created a bimodal frequency for ER+ tumors in which there is a minor peak mainly of luminal B tumors at about age 50 and a major peak due to the late contribution of luminal A tumors near age 70 [3]. This is not seen in Sulaimaniyah for Kurdish or Arab breast cancer patients where the average is < 50 years old for both ethnic groups. In these Iraqi patients, luminal B tumors were the predominant subtype from the beginning of breast cancer development in the early 20s and remained predominant at all later ages. The age distribution of both luminal A and luminal B tumors was unimodal and closely overlapping with no suggestion of a late peak for ER+ tumors.
On the basis of both molecular testing and IHC surrogates, this greater proportion of luminal B over luminal A breast cancers has been observed in other low   Fig. 3 Age specific incidence of the four major ER and HER2 subtypes Legend: Polynomial regression plots show a peak incidence predominantly of ER+/HER2-tumors at approximately age 55-60 years that rapidly declines afterward Fig. 1 Kernal density plots for the four major ER and HER2 subtypes Legend: ER+/HER2+ tumors peak at a significantly earlier age than ER+/HER2-tumors (P < 0.01). There is no significant difference in the age distribution of other subtypes (ANOVA, P > 0.05) incidence countries, including Colombia [age standardized incidence (ASI), 39.7] and China (ASI, 31.7) [17][18][19].
In reports of Middle-Eastern patients in which luminal A were distinguished from ER+/PR+ luminal B tumors on the basis of a low Ki67 index, a > 3:1 predominance of luminal B over luminal A was found in Saudi Arabia (ASI, 24.7) and in Moroccan Arab immigrants to Belgium [20,21]. In the Belgium study, Moroccan and European women were compared [21]. The mean age of Moroccan women was 49 ± 11 years old. European women who averaged 60 years old showed a contrasting all age parity in the proportion of luminal A and B tumors. A 2013 report from Jordan showed that despite an increasing incidence of breast cancer in older women, the median age of all Jordanian breast cancer patients was still 50 years old [22]. The authors claimed a predominance of luminal A tumors, but the cancers were not identified as luminal A based upon the St. Gallen's guidelines and included all luminal subtypes except those that were HER2 positive [22].
The 2013 St Galen's guidelines recommended that luminal A be defined as ER+/PR+/HER2-tumors with a Ki67 index ≤ 14% and PR expression in > 20% of cells [12]. Higher proliferation rate ER+ tumors including the triple positive ER+/PR+/HER2+ subtype are considered luminal B. The 2015 guidelines suggested a Ki67 index ≤ 20% or a Ki67 index less than the laboratory average for ER+/HER2-tumors [13]. Regardless of the specific Ki67 criteria, a low proliferation rate is needed to reasonably replicate a luminal A molecular phenotype.
In Sulaimaniyah patients, triple negative and ER-/ HER2+ were nearly evenly distributed throughout all age groups at a frequency approximately half that of luminal A tumors. This is representative of triple negative and ER-/HER2+ subtypes in other low incidence countries and illustrates their relatively low contribution to the total breast cancer burden compared to luminal tumors [5,18,19,23,24]. Triple negative tumors are unusually common in sub-Saharan and African American women [3,24,25]. However, in other racial groups they seem to have a relatively constant frequency of about 10-15% and increase proportionately across all ages as the general rate of breast cancer rises [3,5,[17][18][19][24][25][26].
The relatively higher rates of luminal B breast cancers in developing countries may be a reflection of the younger populations and the relatively small proportions of aging women exposed to premenopausal hyperestrogenism. We have previously reported that the age specific incidence of the common ER+/HER2-breast cancer is similar in Kurdish and US women under 50 years old [5]. The difference in the two populations is the absence in Kurdistan of the marked increase in the rates of ER+/HER2-tumors in older women that is seen in the US [5]. This lends credibility to a concept of a prolonged population exposure being needed for the development of luminal A tumors. Nevertheless, it is worth mentioning that the age of the population by itself is not the determining factor. China has a huge, aging population but a low breast cancer age standardized rate and low rates of luminal A tumors [17,23]. In addition, low parity, that is considered a risk factor for post-menopausal tumors in the West, has been public policy in China for decades.
Our findings show that breast cancer incidence was essentially the same in the Kurdish region and in the south of Iraq before 2010 with an age standardized rate of approximately 30 per 100,000 women. After 2009, rates increased to 40.0/100,000 women in the south of Iraq and with a smaller change in Sulaimaniyah. This seems to be following the pattern observed over the past decade in Jordan (ASI, 53.8), Lebanon (ASI, 55.4), and several Gulf States including Kuwait (ASI, 48.0) and Qatar (ASI, 45.0) and may be an early phase in the transition toward higher breast cancer rates [2].
It is not obvious whether the increasing age standardized rate of breast cancer in Iraq and other Middle-Eastern countries is a true biological transition or whether it reflects the increased utilization of urban cancer centers [5,19,26,27]. With a more general awareness of breast cancer, financially able women may selectively be traveling for a level of care that is not within the means of many families. In 2011, 60.1% of breast cancer among Jordanian women was diagnosed in Amman where the crude incidence rate was 47.8/ 100,000 [28]. This compared with the outlying Governates where crude incidence varied from 11.7 to 22.7/ 100,000 women. In the Sulaimaniyah Governate, a similar clustering of breast cancer diagnosis was found in central Sulaimaniyah city. This may indicate urban/rural differences in incidence but as likely represents patterns of accessing cancer care. The technology and specialization needed for modern cancer care necessitates its centralization but also has the potential for creating demographic artifacts. With mammography and MRI, central facilities have the ability to detect small and in-situ breast cancer that may not be found by clinical practice outside of major centers. In addition, women taking advantage of these facilities may be more likely to have risk factors associated with post-menopausal disease.
In many if not most Middle-East countries, luminal subtypes represent more than 60% of breast cancers of all ages and more than 70% after age 60 [5, 6, 20-22, 26, 27]. Whether the pattern of increasing rates of breast cancer in older patients is taking a direction that will resemble the West remains to be seen, and it will be valuable to understand the relationships between cancer subtypes in the region. If the Western pattern is followed, age standardized rate should continue to rise, the average age of breast cancer patients should increase, and luminal A should exceed the frequency of luminal B tumors at around 60-70 years of age.

Conclusions
Older Iraqi women do not show the bimodal shift toward higher rates of luminal A breast cancers seen in the West. The modestly increased breast cancer rates being seen in some Middle-Eastern countries may be the result of factors different from those occurring in the West and may not represent a transition to high-risk breast cancer rates. As an epidemiological study, the major limitation is that the design is observational and retrospective and that breast cancer subtyping is lacking in Iraq outside of the Kurdish region.