The effect of anaemia on normal tissue toxicity and survival outcomes in prostate cancer treated with radical radiotherapy and neo-adjuvant androgen deprivation

Abstract

Objective:

It has been established that survival and toxicity outcomes in some cancer types could be influenced by haemoglobin (Hb) levels. This study aims to determine if pre-treatment Hb is associated with late toxicity or survival outcomes in prostate cancer.

Methods:

Data from one Phase III randomised controlled trial and one single arm translational trial were analysed. Patients had localized prostate cancer and received ≥70 Gy radiotherapy and neo-adjuvant androgen deprivation between 1997 and 2013.

Results:

302 males were included. Median follow-up was 6.8 years for toxicity and 10.3 years for survival outcomes. Patients with Hb below the reference range were more likely to experience Grade 2–3 late gastrointestinal toxicity than patients with Hb within the range (p = 0.050). Neither late genitourinary toxicity, erectile function toxicity, prostate-specific antigen relapse free survival nor overall survival of patients were statistically significantly different between groups.

Conclusion:

Anaemia in prostate cancer is found in the minority of patients and is usually mild. Prostate cancer patients undergoing radiotherapy with low Hb were more likely to experience Grade 2–3 late gastrointestinal toxicity.

Advances in knowledge:

This study is one of the first in the published literature to investigate the role of Hb in prostate cancer toxicity and survival. We have found an association between Hb below the reference range and late GI toxicity. Consideration should be given to further investigating patients with iron deficiency anaemia to guide management options and outrule underlying GI pathology before proceeding with radiotherapy treatment.

Introduction

Hypoxia is well a well-recognised feature of cancer cells and is associated with poorer prognosis¹. In hypoxic cells, there is reduced oxygen-fixation of damaging free radicals, reduced proportion of cells in the radiosensitive phase of mitosis, and increased stabilisation of HIF1α which promotes cell survival and angiogenesis.¹ There is strong evidence supporting the use of antihypoxic agents in squamous cell carcinoma of the head and neck, resulting in improved locoregional control and disease specific survival.²

Anaemia has been associated with tumour radioresistance in a range of malignancies, due to its effect on tumour microenvironment hypoxia. This has been shown to be significant in squamous cell carcinoma of the cervix, head and neck, and lung.^3–5 In patients with squamous cell carcinoma of the cervix or head and neck, red blood cell transfusions have been given in anaemic patients with the aim of improving hypoxia, and therefore radiosensitivity. Transfusions have been shown to improve local control in a retrospective study of patients with cervical cancer.⁶ In patients with head and neck cancer, the randomised Danish Head and Neck Cancer Study (DAHANCA-5) describes a univariate prognostic significance of high haemoglobin (Hb) level for local control, disease-specific survival and overall survival; however, transfusion prior to and during treatment did not improve the outcome in patients with low Hb values.^7,8

Similarly, normal tissue tolerance could potentially be affected by changes in haemoglobin. Hypoxia in normal tissues can cause a pronounced fibrotic process after radiation with increased late effects.^9,10 However, normal tissues have better vascular control than the tumour microenvironment and may be less sensitive to changes in Hb.⁹ The mechanism of normal tissue oxygenation is complex, and there is no evidence to define a level of Hb below which normal tissue hypoxia occurs. Evidence for any effect of low Hb on late radiation toxicity is conflicting.^3,11–13

While there is emerging data on the importance of Hb level in some cancers, there is little published data on its’ role in prostate cancer. It has been suggested that prostate cancer survival outcomes and radiotherapy toxicity could be influenced by Hb levels. The Trans-Tasman Radiation Oncology Groups trial (TROG 96.01) presented late radiation toxicity in their 5 year median follow-up of 818 patients with prostate cancer treated with androgen deprivation therapy (ADT) and external beam radiotherapy (EBRT).¹¹ They found that patients who had a higher pre-treatment Hb had reduced late rectal toxicity. This was the first large clinical trial to suggest an association between Hb and rectal toxicity. To the best of our knowledge, no further trials have investigated the link. Another large epidemiological study from Memorial Sloan Kettering has suggested an improved prostate-specific antigen (PSA) relapse-free survival in patients with high risk prostate cancer treated with a statin, possibly due to its antihypoxic and antiangiogenic properties.¹⁴ However, at present, there is not enough evidence for any effect of anaemia or hypoxia on prostate cancer outcomes to support its use as a prognostic factor.

Given the increasing awareness of the impact of Hb in other cancer types and the paucity of evidence for anaemia in prostate cancer radiotherapy outcomes, we aimed to analyse data from two Cancer Trials Ireland (formally ICORG) clinical trials (ICORG 97–01 and 06–15) to investigate the relationship between Hb, late radiation toxicity and survival outcomes.

Methods and materials

Study overview

This was an unplanned retrospective subset analysis of prospective clinical trial data. Data from two prostate cancer clinical trials (ICORG 97–01 and ICORG 06–15) were analysed to determine if pre-treatment Hb was associated with late toxicity and survival outcomes. All patients received ≥70 Gy EBRT and neoadjuvant ADT. The trials were carried out at two Irish institutions between 1997 and 2013. Details of both trials are given below.

Pre-treatment Hb measurement was mandated for one trial (ICORG 06–15), while for the larger trial (ICORG 97–01) the Hb data were collected retrospectively from the patient’s chart.

The Hb normal reference range was 13–18 g dl⁻¹.

We investigated if Hb below the reference range was associated with outcomes of Grade 2 or greater late gastrointestinal (GI) toxicity, late genitourinary (GU) toxicity, erectile dysfunction (ED), PSA relapse-free survival (PRFS) and overall survival.

Both studies were carried out in compliance with the trial protocols, Good Clinical Practice Guidelines and applicable ethical standards.

ICORG 97–01

A Randomised Trial Comparing Short-term (4 months) and Protracted (8 months) Neoadjuvant Hormonal Therapy Prior to Radiation Therapy of High Risk Localized Prostate Cancer.¹⁵

This trial compared the efficacy of 4 vs 8 months of luteinizing hormone-releasing hormone (LHRH) agonist therapy combined with antiandrogen therapy (ADT) prior to radiotherapy, in terms of biochemical relapse-free survival. Patients had localized invasive prostate cancer, stage T1 or T2 (PSA ≥20 or Gleason score at least 7) or Stage T3 (any PSA, any Gleason score).

All patients received triptorelin intramuscular injection once a month and oral flutamide three times daily. All patients then underwent radiotherapy, 74 Gy in 37 fractions. Patients were followed every 3 months after radiotherapy.

GI and GU toxicity were graded using the Radiation Therapy Oncology Group (RTOG)—European Organization for Research and Treatment of Cancer criteria.¹⁶

Erectile function was recorded using a 4-point grading system. ED included any patient who noted a decrease in erectile function from Grades 1–2 (sufficient for penetration) to Grades 3–4 (insufficient for penetration) after treatment.

ICORG 06–15

Proteomic Analysis of Combined Hormonal Therapy and Radiation Therapy for Localized Prostate Cancer.¹⁷

This was a multicentre study using proteomic analysis of sequential serum and urine samples to detect changes in profiles that may predict outcome and identify prognostic and biochemical markers of early disease progression. Patients had intermediate risk localised invasive prostate cancer. They received induction hormones (Bicalutimide and Luteinising hormone-releasing hormone agonist) for approximately 4 months prior to starting radiotherapy. Patients then received radiotherapy consisting of 74 Gy up to a maximum of 81 Gy to the prostate gland.

Late toxicity was graded using the RTOG/European Organization for Research and Treatment of Cancer late morbidity criteria.¹⁶ Erectile dysfunction was graded according to the NCI-CTC toxicity criteria. For the analyses, EF grades were collapsed into two categories, i.e. Grade 1 (sufficient for penetration) and Grades 2–3 (insufficient for penetration).

Statistical methods

Categorical variables were analysed using χ² or Fisher’s exact tests and continuous variables were analysed using independent-samples t-tests. The Kaplan–Meier method¹⁸ was used to estimate survival times. PRFS was as per the Phoenix consensus definition.¹⁹ All statistical tests were two-sided and assessed for significance at the 0.05 level. Statistical analyses were carried out using IBM^® SPSS^® statistical software v. 24.

Results

Patient data

There were 339 patients on the 2 prostate cancer clinical trials in total. Of these, baseline pre-treatment Hb measurements were available for 310 patients. Of these, 302 patients received the prescribed treatment course.

Patient and tumour characteristics are shown in Table 1. Median age at randomisation was 67 years (range 46–81). There was generally an even split between intermediate, high and very high-risk groups. Total ADT duration ranged from 4 to 8 months. The majority of patients received an EBRT dose of 70 Gy in 35 fractions (80.5%, n = 243). The others received 74 Gy in 37 fractions (16.2%, n = 49) or 81 Gy in 45 fractions (3.3%, n = 10).

Table 1.

Patient characteristics

ICORG 97–01 Trial % (n)	80 (243)
ICORG 06–15 Trial % (n)	20 (59)
Age at randomisation (years) Median, Range	67, 46–81
T stage % (n)
T1b-T2a	31(95)
T2b-T2c	17 (51)
T3a	25 (76)
T3b-T4	27 (80)
Gleason score % (n)
<=6	36(109)
7	52(157)
>=8	12(36)
Pre-treatment PSA (ng/mL) % (n)
0–10	33 (99)
10.01–20	35 (106)
20+	32 (97)
Pre-treatment Hb below the reference range % (n)	7 (22)
Risk grouping % (n)
Intermediate	30 (91)
High	43 (131)
V.High	27 (80)
Treatment
AST % (n)
4 months	39 (119)
6 months	11 (33)
8 months	50 (150)
Duration (months) Median, Range	6, 4–8
EBRT
Cumulative dose (Gy) Median, Range	70, 70–81
Fraction size (Gy) Median, Range	2, 2–1.8

AST, androgen-suppression therapy; EBRT, external-beam radiotherapy; Gy, Gray; PSA, prostate-specific antigen.

The median time between pre-treatment Hb to start of androgen-suppression therapy was 0.75 months (range 0–8). The median time from pre-treatment Hb to start of RT was 5.9 months (range 3.4–12.7).

Median Hb was 14.8 g dl⁻¹ (9.7–18.6 g dl⁻¹), with the majority of patients (93%, n = 280) having Hb within the normal range. Those with Hb below the reference range (7%, n = 22) were mostly in the mild anaemia range (median 12.7 g dl⁻¹, range 9.7–12.9 g dl⁻¹), with only two patients having a moderate anaemia (Hb <11.0 g dl⁻¹).

Median radiotherapy dose was 70 Gy in both groups, however the majority of those with a Hb within the reference range received 70 Gy (83%, n = 231) whereas those with Hb below the reference range received either 70 Gy (55%, n = 12) or 74 Gy (41%, n = 9). Less than 5% of patients overall received the higher dose of 81 Gy.

Of those with Hb lower than the reference range, 32% received 4 months ADT and 68% received 6–8 months ADT. Similarly, of those with Hb within the reference range, 40% received 4 months ADT and 60% received 6–8 months ADT.

There was no statistically significant difference in the patient age in those with low Hb vs normal Hb.

Late gastrointestinal and genitourinary toxicity

Late GI and GU toxicity data were available for 97% of patients. Median late toxicity follow-up was 6.8 years (range 4 months–12.2 years).

Over half the patients experienced late GI toxicity to some degree, although the majority of these were Grade 1 (Table 2). There was a statistically significant increased rate of late GI Grade 2 or greater toxicity in patients with Hb below the reference range. 27% of patients (n = 6) with Hb below the reference range experienced Grade 2 or greater late GI toxicity, whereas 12% of patients (n = 32) with Hb within range experienced Grade 2 or greater late GI toxicity (p = 0.050; Fisher’s exact test).

Table 2.

Late GI toxicity

Toxicity	Hb in reference range, n (%)		Total, n (%)
	No	Yes
No toxicity	8 (36%)	125 (46%)	133 (46%)
Grade 1	8 (36%)	113 (42%)	121 (41%)
Grade 2+	6 (27%)	32 (12%)	38 (13%)
Total	22	270	292

GI, gastrointestinal.

There was no statistically significant association between patients with Grade 2 or greater late GU toxicity and whether Hb was in the reference range (Table 3). Over half the patients experienced late GU toxicity to some degree, although the majority of these were Grade 1. Almost 14% of patients (n = 3) with Hb below the reference range experienced Grade 2 or greater late GU toxicity, whereas 17% of patients (n = 46) with Hb within range experienced Grade 2 or greater late GU toxicity (p = 0.871).

Table 3.

Late GU toxicity

Toxicity	Hb in reference range, n (%)		Total, n (%)
	No	Yes
No toxicity	9 (41%)	122 (45%)	131 (45%)
Grade 1	10 (45%)	104 (38%)	114 (39%)
Grade 2+	3 (14%)	46 (17%)	49 (17%)
Total	22	270	294

GU, genitourinary.

There was no statistically significant association between radiotherapy dose and late bladder or bowel toxicity. We have previously published the results of the ICORG 97–01 study, showing no statistically significant difference in late GI or GU toxicities between the 4 month and 8 month ADT arms¹⁵ .

Late ED data

ED data were available for 92% of patients (n = 278). Of these, 65% (n = 181) had erectile function at baseline. The analysis of late ED was based on 180 of these patients (late ED data were unavailable for 1 patient). There was no statistically significant association between patients with late ED and Hb within and below the reference range (Table 4). 75% of patients experienced ED; 85% (n = 11) with Hb below the reference range and 74% (n = 124) with Hb within range (p = 0.523).

Table 4.

Late ED for patients who had initial EF

Erectile dysfunction	Hb in reference range, n (%)		Total, n (%)
	No	Yes
No	2 (15%)	43 (26%)	45 (25%)
Yes	11 (85%)	124 (74%)	135 (75%)
Total	13	167	180

ED, erectile dysfunction; EF, erectile function.

Survival data

Neither overall survival nor PRFS of patients whose pre-treatment Hb was below the reference range were statistically significantly different from those whose Hb was within the range. Median overall follow-up was 10.3 years (range 1.0–19.4). The cumulative proportion surviving at 12 years was 51% for those whose Hb was below the reference range (95% confidence interval: 21–81%) compared to 63% for those whose Hb was in the range (95% confidence interval: 56–69%).

Median PRFS was 7.9 years for those whose Hb was below the reference range (95% confidence interval: 5.5–10.4 years) compared to 7.1 years for those whose Hb was in the range (95% confidence interval: 5.9–8.4 years; p = 0.888).

A Cox proportional hazards model was used to assess the effects of Hb below the reference range in the presence of RT dose, on oncological endpoints. In Cox regression modelling, RT dose was predictive of both overall survival and PRFS and it remained statistically significant in the presence of Hb below the reference range. The Cox regression analysis showed that the hazard is lower and thus the prognosis better in terms of both overall survival and PRFS, for males who received a higher RT dose and this remained the case after adjustment for the effects of Hb below the reference range.

Discussion

Our results from analysis of these two clinical trials show that significant anaemia in prostate cancer is found in the minority of patients. However, we did find a statistically significant increased rate of late GI toxicity in patients with Hb below the reference range. This is similar to the previously published TROG 96.01 findings.¹¹ In a subset analysis of their randomised controlled trial, they found a higher pre-treatment Hb conferred a small but significant protective effect on late proctopathy. This finding is also in keeping with radiobiological principles whereby radiotherapy toxicity could, in theory, be increased by anaemia-induced hypoxia in normal tissue. However, most published literature focussing on low Hb and radiation toxicity is in squamous cell carcinoma of the head and neck and cervix and fails to find an association between the two.^3,12,13

Although we show an association between low Hb and late GI toxicity, the mechanism of this association is unclear. It is unlikely that the mild anaemia seen in our patient cohort could cause normal tissue hypoxia—albeit that little is known about the level of haemoglobin required to cause normal tissue hypoxia, with or without irradiation. It is more likely that the low haemoglobin in these patients could be related to underlying GI pathology, thus associated with increased GI toxicity. In a well-resourced setting such as ours, the most common cause of anaemia in males is occult bleeding. It has been previously reported that male patients with iron deficiency anaemia have a high incidence of GI pathology, both benign and malignant,²⁰ with up one-third of patients having an endoscopic diagnosis of inflammatory bowel disease, high-risk polyp or malignancy.²¹ Consideration should be given to further investigating patients with iron deficiency anaemia to guide management options and outrule underlying GI malignancy.

We found no association between low Hb and late GU toxicity or ED toxicity. Late GU toxicity of Grade 2 or greater has been reported as between 12 and 30% which is in keeping with our findings.^22,23 There was a high proportion of patients with baseline ED prior to treatment (35%), which has been widely described previously.^23–26 We found a rate of late ED of 75% in those without baseline ED. Although methods of evaluating ED vary in the literature, previous prospective non-randomised studies have found similar rates of ED after EBRT, from 59 to 70%.^23,25–27

We also found that overall survival of patients whose pre-treatment Hb was below the reference range was not statistically significantly different from those whose Hb was within the range. Anaemia has been associated with radioresistance in squamous cell carcinoma of the cervix, head and neck and lung, due to tumour hypoxia.^{3–5,13,28–33} However, anaemia in prostate cancer has not yet been proven to effect survival outcomes.^11,34,35 Our finding that low pre-treatment Hb is not associated with survival is in keeping with the available literature. Pai et al investigated the impact of anaemia in 563 patients with localised prostate cancer and found that anaemia was common but was generally mild and was not associated with disease free survival or overall survival; albeit with a relatively short median follow-up of 4.25 years.³⁴ Our median overall follow-up of 10.3 years is a more clinically appropriate follow up for assessing survival in this patient cohort, and confirms that no association has been found.

Our results show that the majority of patients with prostate cancer had a normal Hb, and those with anaemia were almost exclusively Grade 1 anaemia. This is in keeping with previously published literature.³⁴

This study is a subanalysis of data collected from two clinical trials and as such has a number of limitations. The main limitations are the time-lag between Hb measurement and RT, the lack of comorbidities data, and the low number of patients with a low Hb.

The Hb data for the bigger (n = 276) of the two trials was collected retrospectively and was not mandated by the trial. Therefore, the duration of time between Hb measurement and treatment varied—although the Hb measurement was taken at time of randomisation, which was prior to treatment in all cases. Patients also had a varying duration of ADT prior to EBRT, although all received short course ADT, and therefore the time between Hb measurement and RT varied. Hb levels were not routinely re-checked prior to RT so there are insufficient data to suggest any recovery or deterioration of Hb prior to RT.

Another limitation in this study is that comorbidities were not systematically recorded. These may affect the risk of late rectal toxicity and it was not possible to further analyse the impact of conditions such as diabetes, colitis, diverticular disease or vasculopathy on our results. Additionally, the number of patients with Hb below the reference range is quite small (n = 22) and, of those patients, most had mild anaemia, so no conclusion can be reached for patients with moderate–severe anaemia. As with all subset analysis of pooled trial data, these results should be interpreted with caution.

This study is one of the first in the published literature to investigate the role of Hb in prostate cancer toxicity and survival. We have found an association between Hb below the reference range and late GI toxicity, but no association with late GU or ED toxicity or survival outcomes. Further research focusing on the role of Hb in rectal toxicity is required before anaemia can be used as a prognostic factor for late GI toxicity.

Conclusion

Anaemia in prostate cancer is found in the minority of patients and is usually mild if present. Prostate cancer patients undergoing radiotherapy with Hb below the reference range were more likely to experience Grade 2–3 late GI toxicity than patients with a normal Hb. Due to the limitations of this study, no definite conclusion on the effect of anaemia on radiotherapy toxicities can be reached. However, we suggest that consideration should be given to investigating patients with anaemia for underlying GI pathology before proceeding with radiotherapy treatment. The role of Hb in radiotherapy toxicities for prostate cancer remains undefined, and larger more robust trials are needed to confirm the findings of this study.