Regional outcome disparities in German head and neck cancer patients: Shorter survival in Eastern Germany

Abstract Introduction Demographics are important prognostic factors in malignant diseases. A nationwide analysis concerning the prognostic impact of demographics in head and neck cancer (HNC) patients (HNCP) has not been performed previously. Methods A retrospective analysis of data from the Center for Cancer Registry Data (ZfKD) and the Federal Statistical Office (Destatis) between 2002 and 2017 was performed. A total of 212′920 HNCP were included. Incidence, tumor stage, age development, sex distribution, age‐, residence‐, and diagnosis‐time‐specific survival were examined. Results Mean age of HNCP increased more rapidly than in the general population (slope coefficient: 0.29 vs. 0.20; p < 0.0001). Higher age and male sex were associated with a worse prognosis. Whereas overall survival (OS) increased from the early to the later observation period for HNCP <70 years, no OS improvement for HNCP >70 years was found. Furthermore, an OS disadvantage was observed for East Germany compared to West Germany (median 47 vs. 60 months; p < 0.0001). This disparity was associated with a disproportionately high ratio of men in East Germany (men/women: 4.4 vs. 3.1; p < 0.0001) and a lower mean age (61 vs. 63 years; p < 0.0001). In addition to stage, age and sex, residence in East Germany were confirmed as an independent factor for OS in a multivariate analysis. Conclusion Finally, three decades after the German reunion, a survival disadvantage for patients in East Germany still exists. This discrepancy may be a result of socioeconomic disparities.


| INTRODUCTION
According to data from the Global Cancer Observatory (GLOBOCAN) in 2020, there were 921′462 reported cases of head and neck cancer (HNC), resulting in 447′307 deaths worldwide. 1The disease exhibited a male-to-female ratio of 3:1. 1 In addition to well-known risk factors such as smoking, alcohol consumption, and human papillomavirus (HPV) infections, [2][3][4] age plays a significant role in increasing the susceptibility of various tissues to cancer development due to cellular damage accumulation and alterations in the endocrine and immune systems. 5or the purposes of our study, we focused specifically on Germany, a central European country bordered by nine nations.The country's geography varies, with the northern part characterized by low-lying plains and coastal areas along the Baltic and North Seas, while the southern regions are dominated by the Bavarian hills and Alps.Prior to the reunification in 1990, Germany was divided into two entities: the Federal Republic of Germany (BRD) in the West and the German Democratic Republic (DDR) in the East.Consequently, different political, economic, and social systems evolved in the East and West over the years.
Presently, high German smoking rates (27% of men and 19% of women smoked daily in 2019 6 ) and an increasing prevalence of HPV (up to 50%) in oropharyngeal cancers have been observed in East Germany. 7he percentage of elderly and multimorbid patients is rising, resulting in higher healthcare costs throughout Germany. 8Additionally, older patients face a greater risk of cancer-related mortality and may not receive the full standard treatment due to various factors.As a result, they may not benefit from advancements in treatment that have occurred over the past two decades. 9ecently, our research revealed a significant increase in the annual incidence rates and mean age of patients at a tertiary cancer center in Germany, particularly among those aged over 70 years. 10Gender has been recognized as a prognostic factor, 11 and the region of residence may be associated with socioeconomic status, which is also considered an important prognostic factor. 12][15] Thus, the objective of our study was to determine the impact of age, gender, and region of residence on the overall survival (OS) of HNC patients in a comprehensive German nationwide dataset.

| Data collection
We requested anonymized data from the German Center of Cancer Registry ('Zentrum für Krebsregisterdaten' (ZFKD)) of patients newly diagnosed with HNC between 2002 and 2017. 16We received data of 212′920 HNCP including sex, anonymized date of birth, county affiliation, rounded date of diagnosis, TNM (tumor, nodal, metastasis) status, and date of mortality follow-up (East Germany: December 2015, West Germany: December 2017) with the endpoint vital status.Information on Epstein-Barr virus or human papillomavirus was not available.Furthermore, we requested data concerning age and sex of the general German population in between 2002 and 2017 at the central information service of the State Statistical Office Baden-Wuertemberg ('Statistisches Bundesamt, Zentraler Auskunftsdienst'), which afterwards coordinated the data retrieval from the 16 individual countries and their statistical offices.

| Data exclusion
For the analysis of HNC incidence, TNM status and demographics of HNCP we were able to use the major part of the data set (n = 212′920).Nevertheless, sometimes data of TNM status was missing (Table 1, Table S1).For 45.1% of HNCP no UICC stage could be created because of at least partly missing TNM status (mostly M status).The UICC stage calculation was based on the 8th version, which was simplified to four stages without subgroups and in which all oropharyngeal carcinomas were included according to p16-negative classification (because of missing information of HPV status in registry data).It is also of importance that the federal state of Baden-Wuerttemberg started reporting data to the cancer registry beginning in 2009 after reporting of cancer registry data became obligatory, legal requirement (Federal Cancer Registry Data Law; 18.08.2009).A delay in reporting seemed likely to occur in 2016 and 2017.In fact, no HNC reports were received from East Germany in 2017.Therefore, we have not analyzed TNM status, mean age and sex of HNCP development in relation to residence during the years 2016 and 2017.
For survival analyses HNCP with missing data on survival time and status were excluded.In addition, patients marked as deceased, but without a date of death and those alive, but without a date of last follow-up or DCO (death certificate only) cases were excluded for survival analysis too.Hence, the cohort available was reduced to n = 193′877.
T A B L E 1 Cohort characteristics.An overview of the entirety of all included HNCP in terms of their primary sites, stage, gender, mean age, age group and residence is provided.the western and eastern federal states (Figure 1D).However, certain regional differences in the incidence rate can be noted.

| Outcomes in patients diagnosed after 2009 improved in younger patients only
To assess the impact of changes in standard treatment during the first half of the observation period, the cohort was divided by the year of initial diagnosis (period  2B).But no significant difference was detected between the two time periods for old HNCP (≥70 years; period 1: 34.6 months vs. period 2: 34.6 months; p = 0.76; Figure 2C).During the whole observation period, the ratio of advanced to early primary tumor status (T3 + 4 / T1 + 2), N+ to N0 and M1 to M0 of newly diagnosed HNC did not change significantly (Figure S1A).However, HNCP in East Germany consistently presented with higher T status (p < 0.0001) and N status (p < 0.0001), but not M status (p = 0.15).Differences in UICC stage between East and West (Figure S1B) were not observed (p = 0.22).Moreover, the portions of different HNSCC entities differed only marginally, but still significantly (p < 0.0001) by a maximum of 3% in East and West Germany.As expected, UICC stage was strongly associated with prognosis (Figure S1c).Median survival was 127 months (SEM = 1.6;CI: 123.8-130.2) in Stage I, 94 months (SEM = 1.5;CI: 91.0-97.0) in Stage II, 71 months (SEM = 1.2;CI: 68.6-73.4) in Stage III and 29 months (SEM = 0.3; CI: 28.4-29.6) in Stage IV HNCP (p < 0.0001).

| Mean age of HNCP was rising, especially in West Germany
The mean age of all included HNCP (Figure 3A,D) during the observation period was 63.2 years (SD = 11.4).In the   meantime, the age of the general population was between 41.5 and 44.4 years (Figure 3C).The highest mean age was observed in laryngeal cancer patients (65.6 years; SD = 10.8) and the lowest mean age in nasopharyngeal cancer patients (58.5 years; SD = 15.3).Mean age increased significantly over time in all HNC primary tumor sites, whereas the lowest increase was observed in nasopharyngeal cancer patients (Figure 3B).In comparison to the age increment in the general population, mean age among all HNCP was increasing at a significantly higher rate (slope coefficient 0.3 vs. 0.2; p < 0.0001; Figure 3C).The increase in mean age among HNCP was significantly faster in the West than in the East of Germany (slope coefficient: 0.3 vs. 0.2; p < 0.0001; Figure 3C).Consistent with this observation, mean age among HNCP in the East was 2.5 years lower than in the West (61.3 years (SD = 11.2) vs. 63.8 (SD = 11.4);p < 0.0001; Figure 3D).

| Male sex and residence in
East Germany were associated with poor survival The sex ratio (men/women) among HNCP was significantly higher in East than in West Germany (mean ratio: 4.4 vs. 3.3; p < 0.0001; Figure 4A,B).However, the sex ratio has decreased in both parts of Germany evenly (p = 0.95) over time (Figure 4A).Male sex was significantly associated with shorter OS (male: 52 months vs. female: 76 months; p < 0.0001; Figure 4C).Residence in East Germany was also associated with significantly shorter median OS (East Germany: 47 months vs. West Germany: 60 months; p < 0.0001; Figure 4D).

| Multivariate analysis confirmed impact of demographic factors on prognosis
Finally, we performed a multivariate cox regression analysis considering UICC stage, age, sex and residence (East vs. West).All factors in the model were confirmed as independent factors for prognosis (Figure 5).The highest HR was found for advanced UICC Stages III, IV (HR: 2.43; 95% CI: 2.39, 2.49; p < 0.0001), followed by male sex (HR: 1.East Germany for each of the tumor entities studied here (Table S2).

| DISCUSSION
This is the first comprehensive, nation-wide analysis of demographic factors on the prognosis in head and neck cancer.Our study confirms reports about increasing age in HNCP from the United States and parts of Germany. 10,17his poses a challenge for health care systems due to a higher degree of morbidity and mortality associated with treatment 9,[18][19][20] and is associated with a significant financial burden on health care systems. 8Frailty is a predictor of severity for complications in HNCP undergoing surgery 21 and surgery is less often performed in older patients. 9A greater increase in mean age in comparison to the demographic development in the general population underlines the need for age-specific assessments to improve cancer care 22,23 and addresses the described outcome disparities by age.At the same time, the decreasing proportion of younger HNCP may be a result of primary cancer prevention efforts such as anti-smoking and anti-alcohol abuse campaigns. 24Unfortunately, there is no HNC screening tool yet, which is also reflected in the constant tumor burden at initial diagnosis during our observation period.There were significant improvements in the clinical management of HNC during the first half of the observation period (2002-2009) with regard to definitive chemoradiation, [25][26][27][28] adjuvant (chemo-)radiation [29][30][31] and palliative chemotherapy. 32These innovations may explain the increased OS in younger and middle-aged patients.However, most of these innovations consisted of treatment intensification, from which older patients seem not to have had a similar benefit. 33,34This is also in line with earlier (1996-2005) evaluations of the Thuringian cancer registry. 15In older patients, molecular biology, treatment goals and available options may differ significantly from younger patients. 9,35,36Thus, there is a definitive need for clinical trials focusing on treatment of older patients.
An additional prognostic factor was sex, increasingly emphasized by the emerging field of sex medicine, also in oncology. 11,37The ratio of male to female patients has been decreasing over the observation period.Potentially due to higher degrees of smoking and alcohol abuse in male patients, these seem to have a higher risk of death.Primary prevention strategies include prophylactic HPV vaccination which will in the future contribute to a decreasing incidence in both male and female patients, especially if vaccination rates can be improved further (approval in Germany for boys since 2018 and for girls already since 2007 38 ).
An alarming result of our analysis is the outcome disparity between East and West Germany.More than 30 years after the German reunion (joining of the DDR to the BRD: 03.10.1990),such a difference in OS among patients indicates the need for health policy action.Both, a more advanced tumor stage at diagnosis and a higher fraction of male patients seem to contribute to this survival difference, whereas the younger mean age of patients in East Germany could counterbalance these influences.Nevertheless, residence in Eastern Germany was confirmed as an additional, independent factor by multivariate analysis.
Thereupon, persisting gaps in general life expectancy between East and West have been reported, but seem to be slowly vanishing after reunification. 39,40At the beginning of our study in 2002, the difference in life expectancy in the general population amounted to 0.4 years for women and 1.5 years for men. 39A big part of the improved life expectancy in the East since then is considered to be a result of a decline in cardiovascular mortality. 39,40owever, socioeconomic factors also exist that may have a strong impact on HNCP and other cancer patients' mortality.In fact, socioeconomic disadvantages such as differences in education, occupational status and income can lead to differences in experienced health burden (e.g.work related) and resilience.These effects may be amplified by differences in access to health care and personal behavior (nutrition, substance abuse, compliance etc.). 41,42In addition, the healthcare system in the East is particularly dependent on modernization. 43Inequalities in health may then even further increase socioeconomic imbalance. 42In Germany, the mean household income among inhabitants in East Germany is lower and there is a higher rate of unemployment 39 indicating an elevated level of socioeconomic stress which may result in a higher risk for substance abuse. 44Similar difficulties seem to be experienced in the USA, here, data on reduced survival of black patients are available; the reasons are very diverse as well. 457][48] Current efforts to equalize these disparities between East and West Germany need to be continued and expanded.In this manner, the consolidation of a welfare state, facilized health care access, and an enhanced political inclusion of citizens are thought to be helpful. 49In addition, a subsequent extension of this study to urban and rural regions in Germany could provide further insights. 50inally, this study shows the need to enhance cancer data availability and accuracy of documentation.Despite of the mandatory reporting of cancer-related data to cancer registries in Germany since 2009, collected data still have a high prevalence of missing data. 51However, the relations between T, N, M status collected here are comparable to U.S. cancer registry evaluations. 45Efforts to specialize care within dedicated head and neck cancer centers may in the future improve data completeness.However, this constrained data recording is similar to other countries. 52In addition, available data from the statistical offices of the federal were not structured uniformly and represent a significant hurdle for data analysis in public health research.

F I G U R E 1 3 . 4 |
Head and neck cancer incidence and age-specific survival in Germany.(A) Newly diagnosed cases of HNC per 100,000 inhabitants in Germany between the years 2002 and 2017 are displayed with a subdivision into three age-related groups.Incomplete data transmission to the cancer registry is assumed since 2016 (indicated with a dotted, vertical line).(B) The relative, age-group specific proportion of HNC incidence during the observation period is shown.(C) Age-group specific overall survival is presented.(D) The mean number of newly diagnosed HNC per federal state is demonstrated in relation with the resident population numbers from the census in 2011.TNM and stage at diagnosis was constant over time

F I G U R E 2
Comparison of time-specific survival.Age-specific survival over 5 years is displayed for patients diagnosed between 2002 and 2009 in comparison to those diagnosed between 2010 and 2017 for (A) patients <50 years of age at diagnosis, (B) patients > = 50 years, but <70 years of age at diagnosis and (C) patients >70 years of age at diagnosis.

F I G U R E 3
Mean age development by tumor entity and residence.(A) The age distribution of HNCP is displayed as box and violin plots by primary tumor site.(B) Mean age during the observation period is shown by primary site.(C) Mean age of all HNCP per year during the observation period compared to those diagnosed in East or West Germany and age development among the general population.The significance level (p) for the comparisons between slope coefficients is displayed as well.(D) A geographical illustration of the regional mean age of HNCP in Germany for the 16 federal states is shown.
26; 95% CI: 1.24, 1.29; p < 0.0001), residence in East Germany (HR: 1.20; 95% CI: 1.18, 1.22; p < 0.0001) and higher age at diagnosis (HR: 1.03; 95% CI: 1.03, 1.03); p < 0.001).It is worth emphasizing that survival was inferior in the F I G U R E 4 Gender distribution and association with imbalance in survival in Eastern and Western Germany.(A) The gender ratio (men/women) in East and West Germany is displayed over time between 2002 and 2015.The significance level (p) for the comparison between those two slope coefficients is displayed.(B) The mean gender ratio is illustrated geographically for each of the 16 federal states.(C) The overall survival in relation to patient sex is shown with p-value resulting from log rank test.(D) The survival over 10 years by region of residence is shown with the p-value from log rank testing.

F I G U R E 5
Multivariate Cox regression of stage and demographic factors.Hazard ratios with 95% CI (whiskers) for tumor stage, gender, place of residence and age at diagnosis are shown from a multivariate cox regression analysis.A hazard ratio of 1 is indicated by the dotted, vertical line.