The UK Genetic Prostate Cancer Study, Family History of Prostate Cancer and Survival Outcomes Journal Club - Zachary Klaassen
March 23, 2023
Zach Klaassen discusses an article entitled "Family History of Prostate Cancer and Survival Outcomes in the UK Genetic Prostate Cancer Study" in this UroToday Journal Club discussion. The UK Genetic Prostate Cancer Study investigated the association between family history and prostate cancer outcomes, including known prognostic factors, prostate cancer-specific mortality, and all-cause mortality. The primary outcome was all-cause mortality and the association between family history. Family history was defined as present if there was any history of a first, second, or third-degree relative with prostate cancer diagnosed at any age. The results showed that men with a family history of prostate cancer had a higher risk of all-cause and prostate cancer-specific mortality. The study also revealed that the risk of prostate cancer was at least twofold higher for men with a brother or father affected and three to fivefold higher if the first-degree relative had an early onset of the disease.
Biographies:
Zachary Klaassen, MD, MSc, Urologic Oncologist, Assistant Professor Surgery/Urology at the Medical College of Georgia at Augusta University, Georgia Cancer Center
Biographies:
Zachary Klaassen, MD, MSc, Urologic Oncologist, Assistant Professor Surgery/Urology at the Medical College of Georgia at Augusta University, Georgia Cancer Center
Read the Full Video Transcript
Chris Wallis: Hello and welcome to this UroToday Journal Club discussion. Today we're talking about a recent publication entitled Family History of Prostate Cancer and Survival Outcomes in the UK Genetic Prostate Cancer Study. I'm Chris Wallis, Assistant Professor in the Division of Urology at the University of Toronto.
With me today is Zach Klaassen, Assistant Professor in the Division of Urology at the Medical College of Georgia. You can see here the citation for this recent publication in European urology led by Dr. Brook and Dr. Eeles. Prostate cancer risk is disproportionately distributed across the population with a large heritable component and the number of affected relatives changes prostate cancer risk. And for men with a brother or father affected, risk is at least twofold higher than those without family history. And if this first degree relative has an early onset of disease defined as an age of diagnosis of 55 years or younger, the risk rises three to fivefold higher than those without family history.
While many diseases, a large heritable component have a small number of identified underlying genetic risk factors, underlying causes of prostate cancer are polygenic with rare and common variants together explaining about 43% of familial disease in European populations. However, there's no commonly identified and highly prevalent driver of mutations or variants.
While the association between family history of prostate cancer and prostate cancer incidents is well established, the association with clinical and survival outcomes is substantially less clear. There are conflicting data on the association between family history in adverse pathological features and oncologic outcomes, and some of these disparities may be due to the operationalization of family history, which is typically done in a binary fashion.
So the aim of this analysis is to assess the association between family history and prostate cancer outcomes, including known prognostic factors, prostate cancer specific mortality, and all-cause mortality. To do this, the authors leveraged the UK Genetic Prostate Cancer Study, which was started in 1992 and recruits some men across NHS Trust hospitals. In this analysis, patients diagnosed with prostate cancer between 1992 and 2019 were included across three different cohorts. NHS central registers were used to flag patients with a prostate cancer diagnosis.
The authors extracted a substantial volume of important clinical data including date of birth, date of prostate cancer diagnosis and date of death, the mode of detection, tumor staging information, Gleason score, primary treatment approach, vital status, the region, self-reported BMI, self-reported ethnicity, PSA level at diagnosis, as well as all available information for all relatives identified.
Family history was defined as present or family history of prostate cancer was present if there was any history of first, second, or third degree relative with prostate cancer diagnosed at any age. Relative's diagnosis must have been confirmed either by medical records or by self- report. And for all participants, the pedigree was derived.
The primary outcome was all-cause mortality and the association between family history and this outcome was assessed using Kaplan Meier curves, as well as Cox proportional hazards models. A number of co variants were included, including age of diagnosis, mode of detection, PSA, T, N and M stage, Gleason score, primary treatment approach, BMI, year of diagnosis, geographic region, ethnicity and recruitment cohort.
Family history is operationalized in three different ways. As a categorical variable based on the number of first or second degree relatives. That is zero for men without a family history, one or two or more as a categorical variable based on the degree of effect relative i.e. none, first degree or second or greater degree. And as a continuous variable based on the earliest stage of a relative's diagnosis of prostate cancer.
To separate the genetic effects from screening or awareness effects that may occur within a family, the authors assessed the timing of diagnosis between the proband and their relatives and the effect of a family history of breast, ovarian or colorectal cancer on prostate cancer outcomes. All analyses were repeated using prostate cancer mortality rather than all- cause mortality and outcome. And in this case a competing risk regression model was used with non prostate cancer mortality treated as the competing event. And I'll hand it over to Zach who will now walk us through these results.
Zach Klaassen: Thanks so much, Chris for that introduction. So I start off the results here with sort of a cohort summary of these patients. So in this study there was 16,340 indexed patients with prostate cancer, of which 6165, 38%, had at least one first degree relative or second degree relative with a confirmed or unconfirmed diagnosis of prostate cancer. These patients accrued a total of 128,750 person years of follow up over which time 4,380 deaths occurred, which was 27% of the total cohort including 2,961 prostate cancer specific deaths, which was 18% of the cohort. Of note, 1,310 men have been followed without an event for at least 15 years.
This table looks at the continuous distribution of family history by variables of interest. And so looking first at age of diagnosis of the patient, we see that the median age of diagnosis for the men that did not have a family history of prostate cancer was 58. And we see among those that the first degree relative was 59, second degree relative was 57 and a third degree relative was 59. In terms of the age of diagnosis of the youngest relative, we see that the relatives were generally older compared to the patients. So if we see here first degree relative, 67, second degree, 72, third degree, 66, this was older than the age of the patients that were diagnosed. Generally between the patients that had a family history and those that did not have a family history or a degree relative, there was no difference in PSA or in BMI amongst these patients.
This is the survival analysis by family history status and I've highlighted here the outcomes of survival. For patients that did not have a family history of prostate cancer, we see that the all-cause survival among those without a family history was 17.3 years with a five-year survival rate of 83.5%, 10-year survival rate of 70.1% and a 15-year survival of 56.7% and a 20-year survival of 41.8%.
What's notable is that when we break this down amongst men that had relatives with family history of prostate cancer as well as the degree of relative, the patients with no family history of prostate cancer had significantly less survival in terms of the median as well as the 5, 10, 15, and 20 year all-cause mortality compared to those with no family history. So we see that generally to summarize this table that no family history patients had worse survival outcomes.
This is again reiterated in the all-cause mortality among prostate cancer patients by number of affected relatives. We see here that no family history is isolated in blue, one relative is in red and greater than or equal to two relatives is in green. And we see that those with no family history had worse survival compared to those with one or two relatives. Again, this is reiterated when we look at the degree of family history of the affected relatives, no family history again in blue and greater than or equal to a second degree relative or a first degree relative in green and red respectively. We see that the Kaplan Meier curve is significantly worse for those men that had no family history of prostate cancer.
Again, to further reiterate, this is the table looking at risk of overall mortality by number, degree and age of relatives. This top part of this figure looks at risk of death by number of first and second degree relatives with no family history as a reference. We see that compared to one relative, hazard ratio for mortality is 0.85 with a significant confidence interval, the 0.79 to 0.93 for two or more relatives hazard ratio of 0.80 with a 95% confidence interval of 0.71 to 0.90. This is once again seen with a risk of death by degree of relative for first degree relative compared to no family history, 18% risk of reduction in mortality for those with a first degree relative, a hazard ratio of 0.82 and a 95% confidence interval, 0.75 to 0.89. Several other summary points from the results with regards to screening and possible mechanisms for age of diagnosis was strongly correlated with the first degree relative's age of diagnosis. It increased of 0.2 years for every five years increase in the relatives age of diagnosis. However, there was no association for second degree relatives age of diagnosis.
With regards to screen detected prostate cancer. This was associated with a decreased risk of death for those detected from clinical symptoms as ratio of 0.77 irrespective of family history status. Finally, family member affected. We see that for only fathers that were affected, there was a reduction mortality hazard ratio of 0.77 and for only brothers affected, the reduction mortality was 18%, hazard ratio of 0.82. There was no mortality trend when looking at the effect of other cancers such as breast, ovarian, and colorectal cancer.
Secondary outcomes from this study on multi-variable analysis after addressing for family history, no surprise, there was worse overall survival associated with higher T stage, M stage, N stage, higher Gleason score, higher PSA, higher BMI and a more historic diagnosis.
Patients with black ethnicities interestingly were at lower risk of death than those with European ethnicities with a hazard ratio of 0.61 and 95% confidence interval 0.52 to 0.72. Importantly, as we've shown above, these are for overall survival outcomes and there were similar results for prostate cancer specific mortality.
By way of discussion, while a family history of prostate cancer is known to increase the incidence of prostate cancer, the effective family history of prostate cancer on survival outcomes among prostate cancer patients is somewhat contentious. This is a meta-analysis that the authors referenced in their discussion of this article. And this is looking at four of the more prominent studies in this disease space. And we see that the overall hazard ratio on the meta-analysis for these four studies is not statistically significant for family history of prostate cancer and prostate cancer specific mortality with a hazard ratio of 0.93 and a 95% confidence interval of 0.6 to 1.29.
However, we can see that in the current study there is a list that has ratio of 0.85. There's another study by Westerman and colleagues with a hazard ratio of 0.62, showing a protective effect of family history of prostate cancer and prostate cancer specific mortality.
And so although there is no effect when we combine these studies, the authors do mention that the two biggest studies, the Westerman et al study as well as the current study we're discussing today, these were large studies that clearly showed that family history of prostate cancer was protective for prostate cancer specific mortality.
Additionally, this study's ability to refine a definition of family history from a simple yes no to a more informative presentation describing the number and closeness of relatives enabled a more comprehensive assessment of family history compared to previous studies. Increasing the closeness or number of relatives affected is likely to increase awareness of the disease producing a protective effect, which may be similar to screening.
The authors did list two key limitations of the study, which included missing epidemiological risk factors such as smoking and comorbidities, especially when considering an all-cause mortality endpoint. And the second limitation was that this study was predominantly European ancestry at approximately 90% of the patients.
So in conclusion, this study demonstrates a strong inverse relationship between the strength of a family history of prostate cancer and overall survival outcomes in men with prostate cancer. Additionally, this study provides information for the benefit of patients and their relatives on which clinicians can guide these patients based on their familial risk.
And finally, it may provide some reassurance to patients and clinicians that overall survival in prostate cancer patients with a family history of prostate cancer is not adversely affected. And this may provide a platform for further research and to better understanding the impact this could have on targeted risk, adaptive screening and risk stratification algorithms for men with a family history of prostate cancer.
Thank you very much for your attention and we hope we enjoyed this UroToday Journal Club discussion.
Chris Wallis: Hello and welcome to this UroToday Journal Club discussion. Today we're talking about a recent publication entitled Family History of Prostate Cancer and Survival Outcomes in the UK Genetic Prostate Cancer Study. I'm Chris Wallis, Assistant Professor in the Division of Urology at the University of Toronto.
With me today is Zach Klaassen, Assistant Professor in the Division of Urology at the Medical College of Georgia. You can see here the citation for this recent publication in European urology led by Dr. Brook and Dr. Eeles. Prostate cancer risk is disproportionately distributed across the population with a large heritable component and the number of affected relatives changes prostate cancer risk. And for men with a brother or father affected, risk is at least twofold higher than those without family history. And if this first degree relative has an early onset of disease defined as an age of diagnosis of 55 years or younger, the risk rises three to fivefold higher than those without family history.
While many diseases, a large heritable component have a small number of identified underlying genetic risk factors, underlying causes of prostate cancer are polygenic with rare and common variants together explaining about 43% of familial disease in European populations. However, there's no commonly identified and highly prevalent driver of mutations or variants.
While the association between family history of prostate cancer and prostate cancer incidents is well established, the association with clinical and survival outcomes is substantially less clear. There are conflicting data on the association between family history in adverse pathological features and oncologic outcomes, and some of these disparities may be due to the operationalization of family history, which is typically done in a binary fashion.
So the aim of this analysis is to assess the association between family history and prostate cancer outcomes, including known prognostic factors, prostate cancer specific mortality, and all-cause mortality. To do this, the authors leveraged the UK Genetic Prostate Cancer Study, which was started in 1992 and recruits some men across NHS Trust hospitals. In this analysis, patients diagnosed with prostate cancer between 1992 and 2019 were included across three different cohorts. NHS central registers were used to flag patients with a prostate cancer diagnosis.
The authors extracted a substantial volume of important clinical data including date of birth, date of prostate cancer diagnosis and date of death, the mode of detection, tumor staging information, Gleason score, primary treatment approach, vital status, the region, self-reported BMI, self-reported ethnicity, PSA level at diagnosis, as well as all available information for all relatives identified.
Family history was defined as present or family history of prostate cancer was present if there was any history of first, second, or third degree relative with prostate cancer diagnosed at any age. Relative's diagnosis must have been confirmed either by medical records or by self- report. And for all participants, the pedigree was derived.
The primary outcome was all-cause mortality and the association between family history and this outcome was assessed using Kaplan Meier curves, as well as Cox proportional hazards models. A number of co variants were included, including age of diagnosis, mode of detection, PSA, T, N and M stage, Gleason score, primary treatment approach, BMI, year of diagnosis, geographic region, ethnicity and recruitment cohort.
Family history is operationalized in three different ways. As a categorical variable based on the number of first or second degree relatives. That is zero for men without a family history, one or two or more as a categorical variable based on the degree of effect relative i.e. none, first degree or second or greater degree. And as a continuous variable based on the earliest stage of a relative's diagnosis of prostate cancer.
To separate the genetic effects from screening or awareness effects that may occur within a family, the authors assessed the timing of diagnosis between the proband and their relatives and the effect of a family history of breast, ovarian or colorectal cancer on prostate cancer outcomes. All analyses were repeated using prostate cancer mortality rather than all- cause mortality and outcome. And in this case a competing risk regression model was used with non prostate cancer mortality treated as the competing event. And I'll hand it over to Zach who will now walk us through these results.
Zach Klaassen: Thanks so much, Chris for that introduction. So I start off the results here with sort of a cohort summary of these patients. So in this study there was 16,340 indexed patients with prostate cancer, of which 6165, 38%, had at least one first degree relative or second degree relative with a confirmed or unconfirmed diagnosis of prostate cancer. These patients accrued a total of 128,750 person years of follow up over which time 4,380 deaths occurred, which was 27% of the total cohort including 2,961 prostate cancer specific deaths, which was 18% of the cohort. Of note, 1,310 men have been followed without an event for at least 15 years.
This table looks at the continuous distribution of family history by variables of interest. And so looking first at age of diagnosis of the patient, we see that the median age of diagnosis for the men that did not have a family history of prostate cancer was 58. And we see among those that the first degree relative was 59, second degree relative was 57 and a third degree relative was 59. In terms of the age of diagnosis of the youngest relative, we see that the relatives were generally older compared to the patients. So if we see here first degree relative, 67, second degree, 72, third degree, 66, this was older than the age of the patients that were diagnosed. Generally between the patients that had a family history and those that did not have a family history or a degree relative, there was no difference in PSA or in BMI amongst these patients.
This is the survival analysis by family history status and I've highlighted here the outcomes of survival. For patients that did not have a family history of prostate cancer, we see that the all-cause survival among those without a family history was 17.3 years with a five-year survival rate of 83.5%, 10-year survival rate of 70.1% and a 15-year survival of 56.7% and a 20-year survival of 41.8%.
What's notable is that when we break this down amongst men that had relatives with family history of prostate cancer as well as the degree of relative, the patients with no family history of prostate cancer had significantly less survival in terms of the median as well as the 5, 10, 15, and 20 year all-cause mortality compared to those with no family history. So we see that generally to summarize this table that no family history patients had worse survival outcomes.
This is again reiterated in the all-cause mortality among prostate cancer patients by number of affected relatives. We see here that no family history is isolated in blue, one relative is in red and greater than or equal to two relatives is in green. And we see that those with no family history had worse survival compared to those with one or two relatives. Again, this is reiterated when we look at the degree of family history of the affected relatives, no family history again in blue and greater than or equal to a second degree relative or a first degree relative in green and red respectively. We see that the Kaplan Meier curve is significantly worse for those men that had no family history of prostate cancer.
Again, to further reiterate, this is the table looking at risk of overall mortality by number, degree and age of relatives. This top part of this figure looks at risk of death by number of first and second degree relatives with no family history as a reference. We see that compared to one relative, hazard ratio for mortality is 0.85 with a significant confidence interval, the 0.79 to 0.93 for two or more relatives hazard ratio of 0.80 with a 95% confidence interval of 0.71 to 0.90. This is once again seen with a risk of death by degree of relative for first degree relative compared to no family history, 18% risk of reduction in mortality for those with a first degree relative, a hazard ratio of 0.82 and a 95% confidence interval, 0.75 to 0.89. Several other summary points from the results with regards to screening and possible mechanisms for age of diagnosis was strongly correlated with the first degree relative's age of diagnosis. It increased of 0.2 years for every five years increase in the relatives age of diagnosis. However, there was no association for second degree relatives age of diagnosis.
With regards to screen detected prostate cancer. This was associated with a decreased risk of death for those detected from clinical symptoms as ratio of 0.77 irrespective of family history status. Finally, family member affected. We see that for only fathers that were affected, there was a reduction mortality hazard ratio of 0.77 and for only brothers affected, the reduction mortality was 18%, hazard ratio of 0.82. There was no mortality trend when looking at the effect of other cancers such as breast, ovarian, and colorectal cancer.
Secondary outcomes from this study on multi-variable analysis after addressing for family history, no surprise, there was worse overall survival associated with higher T stage, M stage, N stage, higher Gleason score, higher PSA, higher BMI and a more historic diagnosis.
Patients with black ethnicities interestingly were at lower risk of death than those with European ethnicities with a hazard ratio of 0.61 and 95% confidence interval 0.52 to 0.72. Importantly, as we've shown above, these are for overall survival outcomes and there were similar results for prostate cancer specific mortality.
By way of discussion, while a family history of prostate cancer is known to increase the incidence of prostate cancer, the effective family history of prostate cancer on survival outcomes among prostate cancer patients is somewhat contentious. This is a meta-analysis that the authors referenced in their discussion of this article. And this is looking at four of the more prominent studies in this disease space. And we see that the overall hazard ratio on the meta-analysis for these four studies is not statistically significant for family history of prostate cancer and prostate cancer specific mortality with a hazard ratio of 0.93 and a 95% confidence interval of 0.6 to 1.29.
However, we can see that in the current study there is a list that has ratio of 0.85. There's another study by Westerman and colleagues with a hazard ratio of 0.62, showing a protective effect of family history of prostate cancer and prostate cancer specific mortality.
And so although there is no effect when we combine these studies, the authors do mention that the two biggest studies, the Westerman et al study as well as the current study we're discussing today, these were large studies that clearly showed that family history of prostate cancer was protective for prostate cancer specific mortality.
Additionally, this study's ability to refine a definition of family history from a simple yes no to a more informative presentation describing the number and closeness of relatives enabled a more comprehensive assessment of family history compared to previous studies. Increasing the closeness or number of relatives affected is likely to increase awareness of the disease producing a protective effect, which may be similar to screening.
The authors did list two key limitations of the study, which included missing epidemiological risk factors such as smoking and comorbidities, especially when considering an all-cause mortality endpoint. And the second limitation was that this study was predominantly European ancestry at approximately 90% of the patients.
So in conclusion, this study demonstrates a strong inverse relationship between the strength of a family history of prostate cancer and overall survival outcomes in men with prostate cancer. Additionally, this study provides information for the benefit of patients and their relatives on which clinicians can guide these patients based on their familial risk.
And finally, it may provide some reassurance to patients and clinicians that overall survival in prostate cancer patients with a family history of prostate cancer is not adversely affected. And this may provide a platform for further research and to better understanding the impact this could have on targeted risk, adaptive screening and risk stratification algorithms for men with a family history of prostate cancer.
Thank you very much for your attention and we hope we enjoyed this UroToday Journal Club discussion.