Study Type – Aetiology (cohort)
Level of Evidence 2b
Study Type – Aetiology (cohort)
Level of Evidence 2b
• To describe clinical and histopathological characteristics of Finnish familial prostate cancer (PCa) through a detailed analysis of cases in families.
• In total, 202 Finnish families with 617 histopathologically confirmed PCa cases of confirmed genealogy were collected.
• Complete clinical data, including age and prostate-specific antigen (PSA) at diagnosis, stage, grade and primary treatment, were gathered. The mean (range) number of affected men per family was 3 (2–8).
• All the available diagnostic biopsy samples (n= 323) were collected and regraded by the same uropathologist.
• A population-based cohort of 3011 hospital district Pirkanmaa PCa patients was used as a control group.
• The mean (range) year of diagnosis of PCa was 1993 (1962–2006) and the mean (range) age at diagnosis was 68 (43–98 years).
• The median (range) primary PSA level was 12.0 (0.8–11 000) ng/mL. After regrading, the Gleason score was ≤6 in 38%, 7 in 37% and ≥8 in 25% of men.
• The subset of familial PCa men diagnosed after 1995 had higher PSA levels (P= 9.9 × 10−6) and an earlier age of onset (P= 1.7 × 10−6) than men in the control group, although there were no differences in cancer-specific survival.
• We observed an earlier age of onset and higher PSA in familial PCa.
• However, differences between sporadic and familial or hereditary PCa cannot be truly solved until genetic testing of high-risk genes in addition to family history is used to define PCa families.
• We also emphasize that, when histological samples are collected over a longer study period, reanalysis of the samples by the same experienced uropathologist should be considered.
What’s known on the subject? and What does the study add?
Previous hospital- or population-based cross-sectional studies comparing the clinical and histopathological features of hereditary, familial and sporadic PCa either reported weak trends or no differences in features measured except the age of onset.
In present study we observed higher PSA and earlier age of onset in the subset of 257 familial PCa menin Finnish PCa families.
Prostate cancer (PCa) is the most common malignancy in men in many industrialized countries. In Finland, there were 4235 newly-diagnosed PCa cases with the age-adjusted incidence rate of 82.9/100 000 inhabitants and 817 men died of PCa in 2008 . Over the period 1960–2008, the incidence of PCa has increased dramatically, especially during the early 1990s, reaching the highest incidence in 2006 (115/100 000) .
The underlying aetiology of PCa remains poorly understood, with both genetic predisposition and environmental factors likely playing a role. The most prominent suggested risk factors for PCa are age, ethnic background and family history. A large Nordic consortium study of twins estimated that 42% of the risk of developing PCa may be explained by genetic factors . A large number of studies of hereditary PCa have supported the notion that genetics play a role in PCa aetiology and a number of putative PCa susceptibility genes have been identified [3–6].
Even less understood are the clinical features in patients with family history of PCa. Previous hospital-based cross-sectional studies comparing the clinical and histopathological features of hereditary, familial and sporadic PCa, or population-based studies, either reported weak trends or no differences in features measured except the age of onset[7–13]. There are three studies of men with a family history of PCa showing poorer clinical outcomes, including biochemical failure [14–16] and increased mortality  compared to men without a family history of PCa. A number of other studies, however, have reported no differences between outcomes for family history groups [13,17–19].
Hereditary PCa is often defined by the so-called Carter criteria introduced in study from Johns Hopkins showing a Mendelian inheritance of a susceptibility gene. Figure 1 shows three pedigrees describing the Carter criteria of hereditary PCa . However, in previous studies defining familial PCa, there is more variation, meaning the simple clustering of disease in more distant (second-degree) relatives or having at least one affected first-degree relative or a weaker positive family history than hereditary PCa patients [7,8]. Sporadic PCa is most often characterized as having no known first-degree relatives with PCa [8,14].
The Gleason grading system was introduced over 40 years ago but remains one of the most powerful prognostic factors for PCa . Subsequent to its introduction, many aspects of PCa have changed, including PSA testing, TRUS-guided needle biopsy with greater sampling and immunohistochemistry . The system was updated in 2005, addressing trends in practice toward a grading shift, with rare utilization of Gleason patterns 1 and 2 and consensus that Gleason score 2–4 should rarely, if ever, be diagnosed on needle biopsy [21,22].
There are five million Finns who are genetically relatively homogenous, representing an attractive target population for studying genetic susceptibility to cancer . The population-based Finnish cancer registry covers almost all cancers diagnosed over ≈50 years . The Finnish national electronic population registry was established in 1967 and covers all Finns. Information from the different registries can be linked to provide a powerful and accurate tool for studying cancer risk in relatives of patients with malignant disease in an unbiased manner based on histologically confirmed cancer diagnoses. In addition, hospital records are considered to be excellent quality and clinical data can be collected from different hospitals all over Finland.
The present study aimed to collect and evaluate the clinical and histopathological characteristics of familial PCa in 617 males in 202 Finnish PCa families. In addition, to be able to compare Gleason grading over a long study period, all diagnostic samples were collected, reanalyzed and newly graded.
From January 1995, PCa families with two or more affected men among first-degree relatives have been collected in the Laboratory of Cancer Genetics at Institute of Medical Technology. Families have been identified through referrals from physicians, family questionnaires sent to patients, nationwide registry-based searches  and advertisements placed in newspapers, on radio and on television. Subsequently, between 1999 and 2008, additional families have been participating in the study and integrated into the database using the same methods. All families identified with two or more affected relatives with PCa among first-degree relatives were selected for the present study.
Family histories and histologically confirmed cancer diagnoses were verified from hospital records and the cancer registry. Population registration in Finland has traditions dating back to the 16th Century and is considered to be of excellent quality. Population registries are kept by the church parishes and by local authorities for persons not belonging to any official religious community. Since 1964, a centralized, nationwide, computer-based population registry has been run by the national population Registry Centre. This registry reached complete coverage of all Finns on 1 January 1967. The registration in this national population registry is based on unique personal identifiers, which are now used as the main keys in every major person registry, including the Finnish Cancer Registry.
Information collected for all 617 members in these 202 families with PCa included the parents of the index person, the siblings of the index person, the children of the siblings and the children of the index person. These family data were linked with the Finnish Cancer Registry data allowing identification of all cancers cases in these families; calculations for cancer incidence have been reported previously . Figure 2 shows families 15 and 53 as an example of Finnish familial PCa pedigrees.
The detailed clinical characteristics of all PCa cases in the families were collected from hospital records, including the date at diagnosis, age at diagnosis, reason for diagnosis, primary PSA level at the time of diagnosis, follow-up PSA levels, date of biochemical progression, clinical progression, histology, WHO and Gleason grading of tumour samples (biopsy, TURP, prostatectomy), clinical and pathological staging of the primary tumour (T), regional lymph node (N) and distant metastasis (M), primary treatment and follow-up. Table 1 lists the availability of clinical data for all 617 affected males. For comparison between control data, only a subset of PCa cases diagnosed after 1995 was used (257 cases). For survival analysis, follow-up started at cancer diagnosis and finished on 31 December 2009. Overall survival included all deaths and cancer-specific survival included only death to PCa.
|Variable||Data available (n)|
|Reason for diagnosis||533|
|WHO grading, biopsy||477|
|Gleason grading biopsy||204|
|WHO grading prostatectomy||159|
|Gleason grading prostatectomy||102|
|WHO grading TURP||54|
|Gleason grading TURP||9|
|Pathological staging primary tumour (T)||142|
|Pathological staging regional lymph node (N)||144|
|Clinical staging primary tumour (T)||546|
|Clinical staging regional lymph node (N)||547|
|Clinical staging distant metastasis (M)||546|
A population-based cohort from 1996 to 2009 was selected from patients diagnosed with PCa in the Pirkanmaa Hospital District, which serves a population of ≈600 000 inhabitants. Altogether, 3011 men with a mean age of 68 (37–95) years at diagnosis were selected, excluding men belonging to familial PCa group. The primary PSA median value was available from 2988 (99%) of patients. Gleason scores from the diagnostic biopsy samples were available from 2657 (88%) and WHO grade from 2940 (98%) patients. There were 1133 men who had undergone prostatectomy: tumour stage (T) was available from 1062 (94%) and nodal stage (N) was available from 945 (83%) of these.
Because Gleason grading was introduced to Finland in early 1990s, not all patients included in the present study had this information available. In addition, the criteria for Gleason grading have changed remarkably over the last 20 years. We collected diagnostic prostate biopsy samples from all men in families who had it available in archives and re-evaluated the grading. All biopsy samples available (323; 52%) were evaluated by the same uropathologist (P.K.). The same criteria were followed also in the grading of the control group cancers. In Finland, two PCa grading systems are currently in use, a tripartite WHO grading and the Gleason grading system and, from the samples collected, both of the gradings were redefined . Grading was defined using criteria presented in WHO Classification of Tumours . For clinical stage, the same criteria in WHO Classification of Tumours Tumour-Node-Metastasis staging classification were used: primary tumour (T), regional lymph node (N) and distant metastasis (M) . When prostatectomy was performed, pathological TN stage also was given.
Parameters of the data were calculated using the statistical software R, version 2.11.1, available from the Statistical Department of the University of Auckland New Zealand (licensed by GNU General Public Licence 2010). Because the time frame of the diagnosis of PCa in familial samples and control samples was different for comparisons between these two groups, we used only the familial data (257 men) where a diagnosis was made after 1995 (same for the control data). The comparisons of differences in distribution of tumour stage, grade, age at diagnosis and primary PSA were made using the chi-square test or Student’s t-test. For statistical analyses, only Gleason scores in the range 5–8 were compared to allow a sufficient number of cases in each group. Survival curves were constructed by the Kaplan–Meier method. The log-rank test was used for significance testing with respect to differences in survival.
In the subset data of familial PCa men diagnosed after 1995, the mean (sd, range) age of diagnosis was 66 (9.1, 43–89) years and the median (sd, range) year of diagnosis was 2000 (2.8, 1996–2006) compared to those of 68 (8.3, 37–95) and 2002 (3.3, 1996–2009) in the control group. The primary median (QD, range) PSA level was 12 (17, 0.8–4900) ng/mL in the subset family PCa data, whereas, in the control group, it was 9.5 (6.3, 0.1–15 700) ng/mL. Familial PCa men had higher PSA levels than the control group (16.0 vs 9.5 ng/mL, P= 0.065) (outlier values over 200 were excluded) and earlier age of onset (66 vs 68 years, P= 1.7 × 10−6). In a subgroup of men who had undergone prostatectomy, there was no statistically significant difference in T stage (P= 0.40) or M stage (P= 0.34).
The most common reasons for diagnosis were urinary symptoms (n= 313; 52%); second, an elevated PSA level (n= 173; 28%); and, third, cancer found in transurethral resection of the prostate (n= 47; 8%). The primary treatment was most often surgical castration (n= 167; 27%), followed by prostatectomy (n= 149; 27%), chemical castration (n= 68; 11%), radiation therapy (n= 60; 10%), active surveillance (n= 41; 7%), antiandrogen treatment (n= 16; 3%) and bracytherapy (n= 8; 1%). Histologically, 537 (95%) of the males had adenocarcinoma, 28 (4.9%) had unclassified carcinoma and the histology was unknown in 54 (8.4%) of cases. In Table 2, WHO and Gleason grading are listed separately for TURP and prostatectomy patients. Table 3 lists primary tumour stage and Table 4 lists the regional lymph node stage. The reanalyzed familial biopsy samples had a lower Gleason score (P= 0.04) and WHO grade (P= 0.001) than the control group. Tables 5 and 6 show the different WHO and Gleason gradings for original biopsy samples and after regrading. In the control data, 1424 (54%) had a Gleason score ≤ 6, 783 (29%) had a Gleason score of 7 and 450 (17%) had a Gleason score ≥ 8, and 750 (26%) had well, 1792 (60%), moderately and 398 (13%) poorly differentiated cancer, respectively.
|Prostatectomy patients, 135 (89%), n (%)||32 (24)||94 (70)||9 (7)|
|TURP patients, 54 (9%), n (%)||25 (44)||18 (33)||11 (20)|
|Prostatectomy patients, 101 (68%), n (%)||77 (76)||19 (19)||5 (5)|
|TURP patients, 14 (30%), n (%)||10 (71)||0||4 (29)|
|Variable||Primary tumour stage|
|Prostatectomy patients, 142 (95%), n (%)||3 (2)||96 (68)||42 (30)||0|
|Clinical stage, 546 (88%), n (%)||158 (29)||145 (27)||193 (35)||50 (9)|
|Regional lymph node, N1, n (%)|
|Prostatectomy patients||6 (4)|
|All patients||9 (1)|
|Distant metastasis, M1, n (%)|
|All patients||103 (17)|
|Original, n (%)||93 (31)||172 (57)||36 (12)||301 (100)|
|After regrading, n (%)||42 (14)||197 (65)||70 (21)||301 (100)|
|Original, n (%)||5 (3)||26 (9)||32 (11)||45 (28)||34 (21)||14 (9)||3 (2)||1 (1)||160 (100)|
|After regrading, n (%)||0 (0)||2 (1)||9 (6)||66 (41)||52 (33)||19 (12)||10 (6)||2 (1)||160 (100)|
Based on follow-up data available from 405 (65%) patients, 106 (17%) of males had biochemical progression and 143 (23%) had clinical progression (77 distant progression, 66 local progression). In the familial PCa cohort, 64% of men (392) had died during follow-up and 61% (241) of them had died as a result of PCa. In the control cohort, 29% (872) of men had died during follow-up and 31% (267) of them died as a result of PCa, respectively. Patients in the control cohort had a better overall survival compared to men in the subset of the familial PCa cohort (log-rank test, P= 0.02) (Fig. 3a). However, there were no significant differences in cancer-specific survival in the subset of familial PCa men or controls (Fig. 3b).
All together, 323 (52%) diagnostic biopsy samples were reanalyzed and reclassified. The reasons for missing samples are listed in Table 7, with the most common reason being no availability of sample (sample not available for reanalysis, 68%; sample destroyed, 6%; poor quality sample, 1%). The total number of samples where both the original and reanalyzed grading was available was 301 in WHO and 160 in Gleason gradings, respectively. The newly-graded biopsy samples were more often moderately or poorly differentiated (WHO, P= 8.9 × 10−7; Gleason score, 2.9 × 10−8) (Tables 5,6).
|Biopsy sample not found (destroyed, lost, etc.)||227 (75)|
|Diagnosis from TURP||44 (17)|
|Diagnosis from cytological sample||15 (5)|
|Diagnosis from obduction||1 (0.4)|
|Diagnosis from prostatectomy sample||5 (2)|
|Diagnosis from metastasis||2 (1)|
We observed an earlier age of onset and higher primary PSA levels in the subset of 257 familial PCa men in Finnish PCa families, although there was no difference in cancer-specific survival. The strengths of the present study are the unique and large familial PCa data set and the large control cohort of sporadic PCa patients. The family history and histologically confirmed cancer diagnosis were registry-based confirmed. Clinical data were collected from the Finnish Cancer Registry and medical records were obtained from the hospital, where the patient was diagnosed, treated and followed up. In addition, all the available diagnostic biopsy samples from the family data were regraded, making the grade of samples more reliable and more comparable.
The data for present study were collected retrospectively from medical records dating back to 1962. This is a unique characteristic of the present study as a result of the well stored hospital records and registries in Finland. Previous studies have had more recent study periods, with the earliest starting in 1983, although most of them start in the late 1980s or 1990s [9–11,13,14,16,19,28,29]. Medical practice in the diagnosis and treatment of PCa patients has changed remarkably over recent decades, which can lead to some bias in patients being diagnosed with cancer as a result of different criteria (e.g. the stage drift has been observed after the introduction of a PSA test) . In most studies, patients used in the analyses were selected by different clinical criteria, yet no specific genetic alterations can be exclusively used as the selection criteria. Most commonly, a patient group was selected from the hospital patients who undergo prostatectomy, restricting the analysis to a group of men with clinically localized disease [8–10,13,15,29]. In addition, other criteria have been used for patient selection, such as families linked to HPC1, and data have been divided in half according to the number of affected in families . Furthermore, variable criteria have been used to define familial and hereditary cancer: patients confirming one first-degree relative with cancer , Carter criteria [8,12,13] and even more stringent criteria than the Carter criteria . In the present study, we used the criteria of two affected males among the first-degree relatives. The reasons discussed above cause heterogeneity among collected hereditary or familial PCa data in different countries.
In addition to the present study, to our knowledge, the only other study able to use registry confirmed family history, population-based data over a long time period and hospital records not relying on interviews as an information gathering method is that reported by Bratt et al. . Other studies are based on hospital records and various different methods of data collection [8,9,11,13,16,19,29].
Several studies have suggested that men with inherited PCa tend to be diagnosed at a younger age than sporadic cases [7,10,11,28,31]. Studies not supporting the earlier age of onset in familial PCa most often had prostatectomy patients as the selected study population [9,13,14,16,19]. Valeri et al.  reported an age of onset of 65 years to hereditary, 67 years to familial and 71 years to sporadic PCa. The results of the present study support the previous findings of an earlier age of onset in familial PCa men. However, in a population-based study investigating the risk of developing PCa if there is a positive family history, the highest increased risk was found among young patients with malignant disease . These results were later confirmed in a study by Matikainen et al. , where the incidence of PCa for the relatives of PCa patients was higher for relatives of patients diagnosed at an early (<60 years) and late (>80 years) age .
In the present study, the subset of familial PCa patients diagnosed after 1995 had a statistically significantly higher primary PSA than the control cohort, whereas previous studies investigating primary PSA differences among hereditary, familial or sporadic PCa patients most often reported no differences among the groups studied [9,11,13,16,28,29,31]. In the material from the present study, the median PSA of 12 ng/mL was slightly higher compared to previous studies reporting PSA levels of 6.1–9.2 ng/mL [13,19,28,29,31]. The lower PSA levels in previous studies can be explained by the study population mostly consisting of cases of localized disease among men who underwent radical prostatectomy compared to our population-based familial PCa men who also had more advanced disease. At the time when most of the diagnoses in the hereditary PCa were made, the use of PSA was not prevalent in the Finnish population.
Most previous studies have not reported any statistically significant difference in tumour grade and pathological stage between familial or sporadic PCa [8,9,11–13,15,19]. Bratt et al.. reported that 36% of the affected males had local disease, 22% had distant metastases and 46% had intermediate WHO grading. Keetch et al.  and Bastacky et al.  reported that prostatectomy patients with hereditary PCa had a lower Gleason score compared to sporadic and familial carcinomas. However, the lower grade and stage at diagnosis in hereditary/familial cases may reflect differences in screening behaviours compared to the sporadic group. Spangler et al.  reported that men diagnosed before age 60 years with a moderate family history of PCa were more likely to have a higher tumour stage. In an analysis of 74 HPC1 linked North American families by Grönberg et al. , a higher tumour-grade and more advanced stage disease were observed in the HPC1 linked group. In the present study, in the subset of familial PCa men diagnosed after 1995, there were no statistically significant differences in tumour stage or grade between the familial and control data.
Even in survival analysis previous publications have contradictory results. Bratt et al.  reported that there were no differences in survival between hereditary or sporadic PCa. However, some years later, Spangler et al.  reported that men with familial PCa had a poorer survival. Subsequently, Siddiqui et al.  confirmed the findings of Bratt et al. (2002), stating that there were no differences in survival between familial and sporadic PCa. The findings of the present are in accordance with those of Bratt et al.  and Siddiqui et al. , showing no differences in cancer-specific survival between the subset of familial men and sporadic men. However, a limitation of the present study is a loss to follow-up for 35% of the data.
There are some limitations to the present study. First, only the familial histopathological samples were regraded. However, this was carried out by the same experienced uropathologist who had either performed the grading or under whose guidance the grading of the control samples was made. Therefore some intervariability exists among the different pathologists analyzing the control data. Second, over the years, the diagnostics of PCa has improved significantly so that smaller cancers are found. Therefore, we need to have same follow-up duration for familial and control data, leading to a remarkable endpoint in family data.
In present study, we observed higher PSA levels and an earlier age of onset in the subset of 257 familial PCa men in Finnish PCa families. The criteria for selecting hereditary or familial PCa families are not uniform. In addition, the collection of nationwide family data is demanding as a result of the different structures of health care and population registration methods. Because of the late onset of PCa, data collection for a family with three generations needs to account for a family and medical history of >100 years. Because no major PCa susceptibility genes covering most inherited cases have been found, we need to rely on pedigree analysis and patient interviews in the family collection. Patients with true hereditary cancer or the sporadic control group cannot be identified until the heterogenous genetic predisposition of PCa is further revealed. We also emphasize that, when histological samples are collected over a longer study period, a reanalysis of the samples should be considered. In the future, medical decision makers should carefully address the challenges involved in storing medical files, in addition to pathological samples, so that they are able to use population-based study protocols in cancer research with the aim of finding better diagnostic and curative tools for patients.
The authors thank Riitta Vaalavuo for excellent assistance. This study was financially supported by the Competitive Research Funding of the Pirkanmaa Hospital District, the Reino Lahtikari Foundation, the Finnish Cancer Organizations, the Sigrid Juselius Foundation and the Academy of Finland (grant 118413). The Lindell Foundation, Päivikki and Sakari Sohlberg Foundation and The Finnish Medical Society Duodecim also provided financial support to S.P.