Biological ageing and frailty markers in breast cancer patients

Introduction

Chronological age does not always accurately reflect functional status and life expectancy. Frail patients exhibit severely reduced physiological reserves that render them susceptible to minor stressors. More and more oncological therapies become available to older cancer patients, but an accurate evaluation of clinical frailty and general health status is crucial in treatment decision making. Geriatric assessment (GA) is currently the gold standard to evaluate the global health status and clinical frailty level of individuals [1] and is feasible in clinical practice [2], but has several drawbacks. Firstly, it is time-consuming and therefore difficult to integrate in routine clinical practice, although the use of screening tools might partly overcome this problem [3]. Secondly, although GA corresponds with important outcome measures like patient survival and toxicity of treatment, its predictive capacity is moderate, and there is certainly room for better tools. Thirdly, as it does not yield a validated ‘end score’, it is difficult to precisely quantify the patients' global health status. For this reason, some attempts to summarize and categorize GA results have been proposed in geriatric oncology such as the Balducci score, but the included elements and cut-offs are arbitrary, and do not capture the complexity of the entire ageing process (e.g. age ≥ 85 is sufficient to be categorized as ‘frail’, although it has been stated in the geriatric literature that more than half of patients above 85 are actually not frail) [4].

Several biological ageing and frailty markers described in experimental geriatrics have been proposed to reflect ‘biological age’ more accurately than clinical assessment [5], but this still needs to be proven. In fact, it is not always clear whether these biomarkers merely reflect chronological age, or rather the presence of clinical frailty. Anyhow, these ageing/frailty biomarkers have not yet filled the gap from bench to bedside to date. In addition, it should be noted that the oncogeriatric field represents a specific niche where extrapolation of general findings from geriatrics research may not be fully valid.

Telomere length represents one of the best documented markers of ageing. As telomeres were shown to represent some type of cellular ‘mitotic clock’, mean leukocyte telomere length is commonly accepted as a promising ageing biomarker. Indeed, progressive telomere attrition with increasing age has been reported repeatedly. Moreover, shorter telomeres have been linked to age-related disorders such as dementia, cardiovascular diseases, osteoporosis, chronic obstructive lung diseases, cancer, and, most importantly, to a significantly higher mortality rate in the elderly [6]. The association between telomere length and frailty or disability is however much less clear from the literature.

Plasma IL-6 levels have also been associated with mortality and/or worse outcome of several diseases, particularly cardiovascular pathology [7-9]. Several investigators have specifically linked circulating IL-6 levels to the frailty syndrome [10-13]. Notably, IL-6 rising is not an isolated phenomenon but must be seen in the context of a general age-related increase in inflammation markers, called “inflammageing” [14]. Similar age-related differences have also been described for certain inflammatory chemokines such as MCP-1 (also named CCL2, CC-chemokine ligand 2) [15-19], a protein known for its potent ability to attract and activate monocytes/macrophages. Circulating levels of RANTES, another member of the CC-chemokine subfamily, tend to change during ageing [19]. In addition to the above described molecules, certain endocrine markers also show age-related changes. More specifically, IGF-1 is inversely correlated with increasing age [20]. In mice models, disruption of the GH/IGF-1 signaling network resulting in IGF-1 reduction is associated with increase in oxidative stress in the liver, reduced lifespan, and reduced skeletal density [21]. In humans, low IGF-1 levels have been associated with frailty and decreased functionality [22] Although these candidate ageing/frailty biomarkers have been described in the geriatric field, their value in the clinical practice is far from established, and they have not formally been correlated with the frailty syndrome [23] in older cancer patients. The present study was undertaken to validate the five potential ageing/frailty markers mentioned above in a retrospective cohort of older breast cancer patients. In particular, we wanted to investigate the relationship between these markers and calendar age on the one hand, and with the different components of standard geriatric assessment on the other hand.

Results

Discussion

In this study we have investigated the associations between calendar age and clinical ageing/frailty on the one hand, and a panel of biological markers, measurable in blood, on the other hand. In addition, we also describe a novel approach to compile the results of the different geriatric assessment test items into a single output score on a 10-points scale.

Geriatric assessment is currently the best way to assess the level of fitness in oncogeriatric patients, in order to plan an adequate therapeutic strategy. It allows to detect unknown geriatric problems and to adapt treatment regimens accordingly [2, 30-32]. GA consists of different questionnaires and tests that have been validated independently, each reflecting a specific aspect of the geriatric phenotype. However, interpreting GA results and translating them into a risk profile is challenging, since no real ‘end score’ exists. Several frailty models have been developed in general geriatrics, but Balducci and Extermann were the first to suggest a classification of cancer patients into 3 groups (‘fit’, ‘vulnerable’ and ‘frail’) depending on their GA result [33], with the purpose of guiding treatment decisions and predict life expectancy. For instance, frail patients would solely be administered palliative treatments, while a specific individualized approach (e.g. dose reduction at the start with subsequent escalation) would be applied for vulnerable patients. Considering the complexity of ageing, and the wide variety of aspects that are evaluated by GA, the Balducci classification is a simplified tool with many shortcomings and was probably meant as a starting point for future refinement. We have developed a new method, designated LOFS, to calculate a global GA end score that integrates 5 fundamental aspects determining a patient's fitness/frailty status, i.e. capability to autonomously perform activities of daily living (ADL and iADL), mental state (MMSE), nutritional state (MNA-SF) and comorbidities (CCI). A direct comparison between Balducci score and LOFS was not possible, as there is no gold standard to compare them to. However, LOFS is more refined than the Balducci score, as its semi-continuous scoring system is in line with recent evolution in the geriatric research field, where frailty is more and more seen as a continuous event (cumulative deficit model) [4]. In keeping with the Balducci criteria, a patient can be categorized as ‘frail’ solely based on advanced age (85 years or older), or sporadic incontinence. However, patients fulfilling these criteria are not necessarily frail individuals with a global health status being too poor to tolerate cancer chemotherapy. Undertreatment, due to fear of toxicity or intolerance, is a frequent problem in older breast cancer patients and results in an increased risk of relapse [34]. The value of LOFS in predicting toxicity of treatment and general outcome (survival) of patients will be comprehensively assessed in future studies (we are currently conducting a large survival study in a separate cohort).

As GA alone is not sufficient to predict outcome and tolerance to chemotherapy[35, 36], there is an increasing interest to integrate it with specific biomarkers that reflect biological ageing and frailty in individuals affected by cancer, in order to get a more complete picture of the patients' physiological reserves and capability to tolerate chemotherapy. Despite the well-documented correlation of several measurable molecules with age/frailty in the general (non-cancer) population [9-13, 15, 17, 18, 20, 22], none of these markers has emerged as “the” ageing biomarker of choice. Even less information is available on ageing biomarkers that could be valuable in oncogeriatric patients. This is partly due to complexity arising from the extent of the malignant disease, mechanisms of ageing and possible interactions between these two processes. We have explored several candidate ageing biomarkers in our breast cancer study cohort, and examined their relation to the patient's frailty status, either according to Balducci or our newly developed LOFS. Our results show that the biomarkers correlate at least as good with age/frailty defined according to LOFS, as Balducci. In our patient population, plasma IL-6 emerged as the strongest frailty marker. It was not only associated with Balducci category and LOFS, but also with other items of the GA not included in the calculation of these global scores, like for instance ECOG-PS. A correlation between plasma IL-6, ageing/frailty and even with mortality had previously been consistently observed in numerous studies on general (non-cancer) geriatric populations [7-13, 37]. We have shown that plasma IL-6 also correlates with calendar age. Age-related rising of circulating IL-6 is believed to originate from ageing of the immune system, generally referred to as ‘immunosenescence’. This results in an altered profile of circulating leukocytes, but also provokes an imbalance of pro- and anti-inflammatory cytokines and chemokines. This ageing-related pro-inflammatory status, called ‘inflammageing’, has been correlated with dementia, Parkinson's disease, atherosclerosis, type 2 diabetes, sarcopenia, functional disability and death [14, 38]. We now demonstrated that circulating IL-6 levels are associated with both chronological age and frailty in a selected breast cancer population. The additional predictive value of IL-6, supplementary to that of clinical frailty assessment, for outcome and treatment toxicity in (breast) cancer patients should be explored in the future.

For plasma MCP-1 we found a strong correlation with calendar age but no association with clinical frailty. Although association of MCP-1 with atherosclerosis has been documented, circulating MCP-1 levels seem to be primarily correlated with chronological age [15-19]. In line with these earlier reports, our results suggest that MCP-1 merely reflects chronological age, which does not always accurately mirror the patient's physiological reserves and functional status. The lack of association with clinical frailty does not make MCP-1 an attractive biomarker to be used in the clinic for treatment decisions. Data on the chemokine RANTES/CCL5 are even less consistent: RANTES has been proposed as an ageing biomarker because its plasma concentrations have been shown to increase with ageing [39], but in our study we did not find any association with calendar age nor with clinical frailty.

As expected, mean leukocyte telomere length inversely correlated with chronological age in our study population. However, we did not detect any correlation between mean telomere length in circulating white blood cells and functional status. Shorter telomeres have been linked to age related diseases, a higher mortality rate [6], and also to premature ageing syndromes like progeria or trichotiodystrophy [40, 41]. However, studies investigating the relationship between telomere length and frailty/disability showed conflicting results [42-48]. It has been suggested that biomarker associations with health outcomes may differ between very old and younger old populations (through a positive selection effect or survivor effect) [6, 45, 49]. Our population did not contain an overload of ‘very old’, as the median age was 76 years and only 16% of patients were above 80. Nevertheless, we did not find a significant relationship between telomeres and frailty. It should be kept in mind, though, that the oncogeriatric population is probably a biased one, as severely frail older cancer patients will rarely be referred to the oncologists for specialist care. In this regard, observations made on general geriatric populations might not be valid in oncogeriatric populations.

To our knowledge, the association between functional status and mean leukocyte telomere length has never been investigated in older cancer patients; our study is the first to report the lack of such correlation in this specific setting.

The relevance of the IGF-1 pathway in mammalian longevity was initially demonstrated in rodents (calorie restriction was shown to decrease IGF-1 levels with increased lifespan as result) and knockout mice [50]. In humans, low IGF-1 levels have been shown to inversely correlate with increasing age [20], and some studies suggest a link with frailty/functionality [22, 51]. In our study population, IGF-1 showed no association with GA components, except for a significant correlation with CCI. Yet, compared to the other investigated molecules, IGF-1 showed the strongest association with calendar age. Thus, like plasma MCP-1, circulating IGF-1 seems to be linked with the chronological age but not with ageing-related functional decline and frailty.

In conclusion, we have investigated the relationship between several potential biomarkers of ageing/frailty, and the different components of the GA, within a specific breast cancer population. We confirmed IL-6 as a promising marker in predicting frailty, besides its correlation with age. MCP-1, IGF-1 and leukocyte telomere length were correlated with chronological age, but not with frailty, and apparently reflect only the time aspect of ageing, but not the possible functional consequences (i.e. clinical frailty) of the ageing process. However, lack of correlation with frailty status at the time of diagnosis does not necessarily mean that these markers have no value in guiding treatment choices. Hence, the next steps to be undertaken are prospective validation of these and perhaps other markers like for instance circulating microRNAs, in predicting outcome including survival and short- and long term toxicity from different treatment modalities. We are currently conducting an extensive prospective ageing biomarker study in breast cancer patients receiving chemotherapy. The ultimate goal would be the identification of robust frailty biomarkers that can add on, or maybe even (partly) replace, the extensive clinical GA that is suggested nowadays.

We also designed the LOFS, a novel comprehensive method to categorize patients on the basis of their GA results. This new 10-point scoring system reflects the general condition of a patient in a more refined way and allows a more subtle interpretation of the GA results taking into account 5 crucial domains of GA, while at the same time retaining the simplicity of a single end score that is desirable for application in daily practice.

The field of medicine in general, but especially geriatric oncology, is evolving more and more towards a patient-tailored approach in which accurate frailty assessment instruments, such as easy-to-measure biomarkers and reliable but simple clinical evaluation tools, could be very helpful.

Materials and Methods

Supplementary Materials

Supplemental File