Retrospective Analysis of Prognostic Factors of Localised or Locally Advanced Prostate Cancer and Their In uence on Treatment Choice

Sigita Liutkauskiene Oncology Institute of Lithuanian University of Health Sciences Karolina Martinaityte Medical Academy, Faculty of Medicine, Lithuanian University of Health Sciences Rasa Malonyte (  rasa.malonyte@gmail.com ) Hospital of Lithuanian University of Health Sciences Kaunas Clinics https://orcid.org/0000-0003-42746743 Kristina Jureniene Cardiology Institute of Lithuanian University of Health Sciences Saulius Grizas Clinic of Surgery of Lithuanian University of Health Sciences Elona Juozaityte Oncology Institute of Lithuanian University of Health Sciences


Background
Radical prostatectomy and radical radiotherapy are the main treatment options for non-metastatic prostate cancer. Which one of these two methods is more effective still remains the subject of discussion and debate in multidisciplinary team meetings, especially between urologists and radiation oncologists.
According to the majority of retrospective analyses and meta-analyses published in recent years, radical prostatectomy is a more effective treatment method than radical radiotherapy (1,2,3), while radical radiotherapy has been even identi ed as a factor associated with increased overall mortality in some articles (3). However, international guidelines currently recommend radical prostatectomy and radical radiotherapy as level 1 evidence treatment options without distinguishing any method as more effective (4,5,6). Why do the ndings of retrospective meta-analyses differ from international recommendations?
In search of an answer to this question, we decided to conduct a retrospective study to analyse the impact of prostate cancer-dependent prognostic factors and patient-dependent prognostic factors on 5-year overall survival of localised or locally advanced prostate cancer patients and the in uence of these factors on radical treatment choice.

Study design and setting
This retrospective study was approved by the Bioethics Center of Lithuanian University of Health Sciences. Medical records of patients diagnosed with non-metastatic prostate cancer in 2008-2012 in the A liate of Hospital of Lithuanian University of Health Sciences Kaunas Clinics Kaunas Oncology Hospital were analysed.

Participants
Inclusion criteria of the study were the following: 1) histologically con rmed prostate adenocarcinoma; 2) patients treated in the A liate of Hospital of Lithuanian University of Health Sciences Kaunas Clinics Kaunas Oncology Hospital; 3) localised or locally advanced prostate cancer diagnosed in 2008-2012; 4) no distant metastases at the time of initial diagnosis; 5) medical records maintain comprehensive data on treatment and follow-up. Exclusion criteria were the following: 1) medical records maintain incomplete data on treatment and follow-up; 2) history of other previous malignancies; 3) distant metastases suspected at the time of initial diagnosis.
The following data necessary for this study were collected: date of diagnosis, patients age at the time of diagnosis, comorbid conditions, baseline prostate speci c antigen (PSA) level, Gleason score, clinical T category (cT) at the time of diagnosis, method of treatment used, pathological T category (pT) after radical prostatectomy, data on the radicality of the operation (R), date of biochemical recurrence, date of occurrence of metastases, date of last visit, date of death. The date of diagnosis was de ned as the date when pathology report of histological examination of tumour biopsy was provided. The clinical value of prognostic factors and the effectiveness of the treatment method were assessed by analysing the 5-year overall survival of the patients from the date of histological con rmation of the diagnosis.
Three-dimentional conformal external beam radiotherapy was used for irradiation of prostate cancer patients. Dose-escalated intensity-modulated radiation therapy (IMRT) and image-guided RT were not available at that period in our hospital. Daily fractions of 2 Gy, ve fractions per week till 60-70 Gy was standard of care at that time.
The impact of patients' comorbidities on survival was assessed using the age-adjusted Charlson Comorbidity Index (aa-CCI) (7,8). The components of the aa-CCI and their values are shown in Table 1. Acquired immunode ciency syndrome (AIDS) 6 *was not counted because all included patients met this criterion **patients with metastatic disease were excluded from this study

Statistical Methods
The major goal of the study was to evaluate the in uence of prostate cancer-dependent and patientdependent prognostic factors on the choice of radical treatment method. To achieve this goal, we had to assess the impact of prognostic factors and treatment methods on 5-year overall survival. Overall survival at 5 years was de ned as being alive 5 years after cancer diagnosis (the date of prostate cancer histological examination), with those alive censored at the last known follow up. The Kaplan-Meier method was used for single variable survival data analysis. Differences of the survival rate were determined using the log-rank test.
The impact of various factors on the 5-year overall survival of prostate cancer patients was assessed using univariate and multivariate Cox's regression analysis. Hazard ratios (HR) and 95% con dence intervals were calculated.
The in uence of prognostic factors on the choice of treatment method was assessed by making comparisons between groups using the Chi-square test.
Level of signi cance for statistical tests was 0.05.

Microsoft O ce Excel 2007 and Statistical Package for the Social Sciences (SPSS) version 25 were used
for data processing and statistical analysis.

Results
Data of 301 patients who met the inclusion criteria for the study were analysed. For all patients 5-year overall survival was 76.6%. During the 5-year follow-up period 70 patients (23.4%) died and 231 survived for 5 years or more. Distant metastases were diagnosed in 50 patients (16.6%) over a 5-year period.
The process of patients inclusion in the study is shown in Fig. 1.
Characteristics and 5-year overall survival (%) of the study patients, taking into account prostate cancerdependent and patient-dependent prognostic factors, are presented in Table 2. Using Kaplan-Meier method, we found that 5-year overall survival in patients with localised or locally advanced prostate cancer is impacted by patient age, pathological T category (pT), baseline PSA level, Gleason score and the score of age-adjusted Charlson Comorbidity Index (Table 2) Analysis of the impact of patient-dependent prognostic factors on 5-year overall survival Kaplan-Meier method was used to analyse the impact of patients age and comorbidities on 5-year overall survival.
It was proved that the patients age at the time of diagnosis had an impact on the 5-year overall survival (p = 0.024). Comparing survival among age groups, patients aged 60 years and younger had a higher 5year overall survival rate compared with patients aged 75 years and older (p = 0.003) and patients aged 61-65 years had a higher 5-year overall survival rate compared with patients aged 75 years and older (p = 0.046) as well. No impact of age on survival was found in the other age groups (Table 2).
Patients with a higher score of aa-CCI had a signi cant deterioration in 5-year overall survival (p < 0.001).
There was no signi cant difference in survival rates between patients with aa-CCI score of 7-8 points compared with patients with aa-CCI score of 9 points or more (p = 0.113) ( Analysis of the impact of prognostic factors and treatment method on 5-year overall survival 5-year overall survival of patients treated with radical prostatectomy was 87% and 70,9% of patients treated with radical radiotherapy (p = 0.002). Based on this result, RP appears to lead to more favorable patients outcomes and prognosis. This result is consistent with the results of previous meta-analyses (1,2,3).
To identify all factors that determine the survival of patients with localised or locally advanced prostate cancer, we conducted an univariate and multivariate Cox regression analysis to assess the impact of patient-dependent and prostate-cancer-dependent prognostic factors along with treatment method on survival.
Using Cox regression analysis model, we analysed the impact of patients age, baseline PSA, comorbidities, Gleason score, pathological T category and treatment method on 5-year overall survival. Multivariate Cox regression analysis showed 2.54 times higher death risk over 5 years in patients with Univariate analysis showed that patients treated with radical RT had 2.34 times higher death risk (p = 0.003) and patients treated with hormone therapy alone had 3.13 times higher death risk (p < 0.001) compared with patients for whom an operation was performed. Multivariate analysis did not demonstrate a negative impact of RT on survival compared to RP (p = 0.286), but patients who received hormone therapy alone had 2.49 times higher death risk than patients who underwent surgery (p = 0.023). Table 3. In order to nd out the criteria that determine the choice of treatment option in daily clinical practice, we analysed the in uence of patient-dependent prognostic factors (age, the score of age-adjusted Charlson Comorbidity Index) and prostate cancer-dependent prognostic factors (Gleason score, baseline PSA) on treatment choice.
We found that RP was more common in patients with better prognostic 5-year overall survival according to Kaplan Meier method, it means that RP was statistically signi cant chosen in patients younger than 60 years old (p < 0.001), with a lower score of age-adjusted Charlson Comorbidity Index (p < 0.001) and with lower baseline PSA level (p < 0.001) (Fig. 2).
More favorable prognostic factors were found in the RP treatment group compared to the RT treatment group: patients aged 60 years and younger accounted for 35% in the RP group, and 18% -in the RT group, with increasing age the number of RP decreased and the number of RT increased (p < 0.001); score of 3-4 points of aa-CCI was set in 36% patients in the RP group, and 6% -in the RT group, with increasing aa-CCI score, the number of RP decreased and the number of RT increased (p < 0.001); baseline PSA level of 10 ng/ml and less was found in 67% patients in the RP group, and 52% -in the RT group, with increasing PSA level, the number of RP decreased and the number of RT increased (p < 0.001) (Fig. 2).
Gleason score did not in uence the choice of treatment method (p = 0.24).

Discussion
According to many retrospective meta-analyses, radical prostatectomy results in better overall survival rates compared to radical radiotherapy (1,2,3,4). The ndings of the studies were presented as undisputed and recommended prioritising RP for all patients with localised or locally advanced prostate cancer. However, these meta-analyses have their limitations. For example, Walis et al. did not analyse the impact of patient-dependent and prostate cancer-dependent prognostic factors on the survival rates and the in uence of these factors on the choice of treatment method. In the study of Sooriakumaran et al. it was analysed the impact of RP and RT on survival according to patients' risk groups, but there was no analysis of patient-dependent risk factors as well as no analysis of the in uence of prognostic factors on treatment choice. It was clearly demonstrated that the two patients' groups (treated with RP or RT) in this study are de nitely different. In our study it was shown that almost all clinico-pathological data signi cantly favored surgery. In fact, most of the patients treated with RT were older, with more comorbidities and worse features of prostate cancer (eg, higher Gleason score and higher baseline PSA levels) than surgically treated patients.
In our study using a multivariate Cox regression analysis model, we found that Gleason score and patients' comorbidities had a signi cant impact on 5-year overall survival in localised or locally advanced prostate cancer patients. Patients who received hormone therapy alone had 2.49 times higher death risk compared with patients who underwent radical surgery. Can we say that hormone therapy increases the risk of death compared to surgical treatment? De nitely not, as evidenced by the data presented in Fig. 2, which reveals that hormone therapy alone was given to patients with aa-CCI score of 7 or more, 64% patients had baseline PSA greater than 20 ng/ml, 80% patients were 66 years and older -that is, hormone therapy was given to patients with poorer prognosis.
In our study, patients treated with RP and RT had a 5-year overall survival of 87% and 70.9%, respectively (p = 0.002). However, after analysing the distribution of prognostic factors in the treatment groups, we found that patients with better prognosis (younger than 60 years (p < 0.001), with a lower score of aa-CCI (p < 0.001) and lower baseline PSA (p < 0.001) were more likely to be treated with RP (Fig. 2).
Therefore, based on the data of our study, we conclude that it is not the local treatment method that affects survival in prostate cancer patients, but the patient-dependent and disease-dependent prognostic factors which in uence the choice of treatment method.
Another important aspect we would like to highlight is the need for RT after RP. RT after RP is commonly used after non-radical (R1) surgery (about 20% of cases in our study) and after the development of biochemical recurrence (biochemical recurrence after RP occurred in 27% patients and after RT -in 8.7% patients). How should we assess this fact? Should we just know that for about 45% of patients RT will be necessary after RP? Or perhaps the combination of these two treatment methods improves patients' prognosis and survival rates, as it was found in the study by Fahmy et al. (9)? There are many examples in oncology where a combination of these two treatment methods results in better treatment outcomes (breast cancer, colorectal cancer, etc.).
Could prospective randomised clinical trials today answer the question which one of these two treatment methods is more effective? There is no doubt that this is the only way to evaluate the possible superiority of one treatment method over another, but it requires new radiological technologies, experienced physicians, accurate methods to assess the stage of prostate cancer, molecular markers, as prostate cancer is a very heterogeneous disease with a very heterogeneous patients' population (10,11).

Conclusions
Results of our study showed that the survival of localised or locally advanced prostate cancer patients was determined by patient-dependent and disease-dependent prognostic factors which de nitely in uenced the choice of treatment method.  Flow chart of cases used for analysis