Palliative Radiotherapy in Cancer Patients with Increased Serum C-Reactive Protein Level

Discussion

The present study is the first large-scale analysis of the potential role of CRP when estimating prognosis of patients referred to palliative radiotherapy. It has previously been shown that radiotherapy causes inflammation and increases CRP (14-16). Other studies suggested correlations between CRP and prognosis after curative radiotherapy, e.g. in prostate (17), rectal (18) and esophageal cancer (19). Such results are in line with those from studies of other treatment modalities and malignancies (6, 7, 9, 10, 20). The strengths of our study include the large number of patients (statistical power), long follow-up, consistency of pre-radiotherapy assessments, use of only one laboratory and assay for CRP measurement, as well as availability of information about several potential confounders, e.g. comorbidity, steroid medication, intravenous antibiotic therapy and WBC. Disadvantages include the retrospective design and lack of consistent documentation of oral antibiotics. We also acknowledge that a number of different individual physicians provided clinical care to our patients. Their attitude towards initiation of antibiotic therapy might have differed. A common challenge in clinical practice is to differentiate between pure cancer-related CRP increase and concomitant infection. Parameters such as fever, clinical symptoms, WBC, blood and urine culture might point towards infection and support decision-making. Likely, the threshold to start antibiotics was lower in patients with reduced PS, ongoing systemic therapy or other factors indicating that untreated infections could become life-threatening or compromise the oncological treatment plan. The timing of palliative radiotherapy in patients with recently treated infection might also differ. For example, patients with metastatic spinal cord compression might require immediate treatment, while patients with uncomplicated bone metastases can postpone radiotherapy and might have normal CRP before start. It is also important to be aware of the fact that CRP level might change from one day to another.

Our results showed limited impact of steroid medication on CRP level. This finding was unexpected and potentially related to the fact that many patients were in the terminal stage of their disease with known progression outside of the target volume(s) and no further systemic options. In this setting, the influence of steroids might become less pronounced. Certain cancer types and patterns of spread were associated with higher CRP levels, e.g. NSCLC, lung metastases and pleural affection. Patients with high CRP had higher likelihood of anemia, abnormal WBC and reduced PS. Especially, PS is a strong prognostic factor in all established models predicting survival after palliative radiotherapy (12). Our multivariate analysis suggested that CRP is not just a surrogate marker of reduced PS or other adverse prognostic factors. Rather CRP contributed important and complementary information. It would certainly be possible to develop a score that includes the other independent prognostic factors (PS, progressive disease outside of the target volume(s), pleural affection, short time interval from diagnosis to radiotherapy, adrenal gland metastases, comorbidity). However, as shown in Figure 1, CRP alone is sufficient to create a clinically useful 4-tiered score, which is readily available and easy to assess. No expensive analyses or imaging is needed, an important aspect in many countries with limited health care resources.

Use of intravenous antibiotics was not uncommon in patients with high CRP. This is understandable given the potential for serious complications of uncontrolled infection in patients receiving radiotherapy (many patients were elderly, with comorbidity and/or reduced PS). The retrospective design of our study is not adequate to better understand the necessity for antibiotics and their benefits and side-effects in the clinical setting of palliative radiotherapy. We were worried about the short survival of patients with CRP >60 mg/l who had received antibiotics during the two-week time period before radiotherapy (Figure 4). Aggravation or relapse of infection promoted by inappropriately early radiotherapy could be a potential explanation for short survival. However, only one patient (10%) in this small subgroup actually died from uncontrolled infection. Rapid cancer progression was the prevailing cause of death. Most likely these patients had cancer-related CRP increase and the clinicians decided to terminate the antibiotics after a trial period where CRP did not decrease. The complicated issue of optimal utilization of antibiotics in patients with incurable cancer requires prospective studies.

In conclusion, our data provide a strong impetus for further research about CRP and inflammatory markers in general in patients who receive palliative radiotherapy. Even larger databases than ours are needed to optimize and validate our prognostic score, which was based on arbitrarily defined levels of mildly and moderately increased CRP. Biomarkers might improve current prognostic models and facilitate individualized cancer treatment, e.g. in the context of too aggressive or lengthy treatment schedules in poor prognosis patients or near the end of life.