Variability in CRP, regulatory T cells and effector T cells over time in gynaecological cancer patients: a study of potential oscillatory behaviour and correlations

The link between inflammation and tumour development has been widely reported and recently, in an update to their seminal paper on the hallmarks of cancer, Hanahan and Weinberg refer to inflammation as an “enabling characteristic” for tumour development [1]. This link is well illustrated in gynaecological malignancies. The processes of ovulation and menstruation are associated with inflammation of the ovarian surface and endometrial epithelia [2, 3]. During ovulation there is an increase in the macrophage population in the perifollicular stroma of pre-ovulatory follicles [4, 5]. These macrophages produce pro-inflammatory cytokines Interleukin (IL)-6, IL-1β and Tumour Necrosis Factor (TNF), [2]. Oestrogen in the endometrium also induces the up-regulation of these pro-inflammatory cytokines [6]. Over time, the cyclical process of epithelial damage and repair, involving inflammation, predisposes to neoplastic growth. Not only does inflammation play a role in tumourigenesis, it also facilitates disease progression. The tumour microenvironment is rich in inflammatory mediators produced by tumour cells and infiltrating leukocytes [7‐9]. Specifically, IL-6 triggers tumour cells to produce matrix metalloproteinase (MMP)-9 that promotes tumour growth by initiating angiogenesis [10].

C reactive protein (CRP), an acute phase protein produced by hepatocytes is the most widely used indicator of inflammation. The transcription of CRP is induced by IL-6 alone while the presence of either IL-1β or TNF augments its production [11, 12]. Gynaecological cancer patients exhibit elevated levels of CRP, however these levels are not static. In a preliminary study of melanoma and ovarian cancer patients, serum CRP levels oscillated about a mean with a periodicity (λ) of 7 days [13]. The dynamic state of inflammation could indicate an underlying persistent but regulated anti-tumour immune response in the cancer patients, which is characterised by a cyclical repeating process of endogenous auto-vaccination, preceding suppression by regulatory cells.

In the past few decades the role of the adaptive immune system in driving anti-tumour immunity has been highlighted. Studies have shown a positive correlation between the number of tumour infiltrating effector T cells and patient survival, as well as a negative correlation with Foxp3⁺CD25^HiCD4⁺ regulatory T cells [14, 15]. Inflammatory cytokines and chemokines such as TNF and CCL22 facilitate the infiltration of T cells into the tumour microenvironment. When activated, effector T cells up-regulate CD25 and transiently express FoxP3 [16]. This enables them to expand and secrete pro-inflammatory cytokines. FoxP3⁺CD25^HiCD4⁺ regulatory T cells, which express high levels of the CCL22 receptor CCR4, also infiltrate tumour sites where they expand in response to TNF and IL-2 produced by effector T cells, and subsequently inhibit the immune response [17‐19].

A lag between the expansion phase of effector T cell and regulatory T cell populations could account for the oscillations in CRP levels observed in cancer patients. If this is the case, oscillations in serum CRP levels could be used to target regulatory T cells for depletion during their expansion phase, by using chemotherapy. Pre-treatment results are reported herein on a prospective phase II trial of cyclophosphamide treatment for gynaecological malignancy, where repeated sampling was taken prior to trial initiation. This allowed the opportunity to assess the hypothesis that CRP levels oscillate, and may be indicative of effector and regulatory T cell proportions and that the latter may also display predictable oscillations in patients with advanced gynaecological malignancies.

The dynamic nature of the adaptive immune responses has been described in two ways. Firstly, there are diurnal fluctuations, regulated by the circadian clock exemplified in CD4 T cells by the rhythmic expression of genes that control cytokine secretion and cell function [22]. Secondly, antigen dependent fluctuations occur during acute infections and the kinetics of the immune response are controlled by a system of positive and negative feedback mechanisms designed to limit the immune response when pathogenic insults are resolved. In chronic diseases such as cancer, antigenic clearance does not occur and the persistent antigen exposure results in a constant state of immune activation. The tumour however, limits immune activation by secreting immune inhibitory cytokines such as IL-10, and tumour growth factor (TGF)-β as well as inducing regulatory cell populations including myeloid derived suppressor cells (MDSCs) and regulatory T cells [18, 23‐25]. Although intratumoural immune cells are skewed towards immune inhibition, there may still exist a homeostatic balance that needs to be maintained in the periphery of cancer patients in which an oscillating sequence of immune activation is followed by negative feedback immune suppression [13]. The study herein however, showed that CRP concentrations in all the patients did not appear to oscillate periodically nor did the concentration of CRP correlate significantly with T_reg or T_eff frequencies. Therefore, CRP may not be a practical useful surrogate marker for either cell population. In a study of 12 patients with melanoma, less than 50% showed possible time dependent CRP concentration profiles [26]. The two data sets available therefore suggest that inflammation in cancer patients may not consistently follow an oscillatory, sinusoidal (or other mathematical) pattern with a constant period or amplitude. This may be because any such model oversimplifies the inflammatory process, and correlations are easily disrupted by a number of potential additional in vivo co-parameters or process variations. Multiple factors influence the levels of cytokines that regulate inflammation. One such factor is the tumour growth. In patients with papillary thyroid cancer CD4⁺ T cell frequencies correlate with tumour size while T_reg frequencies correlate with lymph node metastasis [27]. Such variables may hence significantly influence the magnitude of T cell effector and suppressor frequency and function.

In our study, neither the coefficient of variation of both T_regs and T_effs nor that of the ratio of T_eff to T_regs, in cancer patients was greater than that of healthy donors suggesting that in the periphery the fluctuation of T_reg within the CD4 population was not affected by the presence of the tumour. The coefficient of variation of T_effs was lower in cancer patients compared to healthy donors, which may have been due to immune suppression. Consistent with this suggestion, the frequency of T_effs was significantly lower in the patients than in healthy donors. Although T_reg frequencies in the periphery were not significantly increased, other cell subsets such as MDSCs may also facilitate immune suppression. Pro-inflammatory mediators have also been suggested to promote accumulation of MDSC in cancer patients’ peripheral blood [28]. Within the tumour microenvironment, the fluxes between T_reg and T_eff populations may be more evident.

Although changes in the frequencies of conventional T_reg and T_eff did not correlate with inflammation, it is still possible that minor subsets within each phenotype (T_reg or T_eff) or their specific function over time, may correlate. Indeed, effector and regulatory T cells are heterogeneous populations of cells. Further breaking down these populations based on phenotype and function may also show greater variation between patients and healthy donors. For example, T_reg populations with enhanced suppressive function due to elevated expression of inhibitory receptors such as glucocorticoid induced tumour necrosis factor receptor (GITR) and cytotoxic T-lymphocyte antigen (CTLA)-4 as well as increased production of suppressive factors such as adenosine and cytokines TGF-β and IL-10, have been reported to be elevated in cancer patients [29‐31]. Similarly, effector T cells can be broken down into different functional phenotypes such as IL-17 secreting and type-1 interferon secreting T_effs. Type-1 interferon secreting T_effs promote proliferation of cytotoxic CD8 T cells, which contribute towards an anti-tumour effect [32‐35]. It cannot therefore be excluded that the frequencies of some of these functional subsets, as well as CD8 T cells, may be subject to regulation by inflammatory factors, even when the results presented herein show that total T_reg and T_eff populations are not correlated to inflammatory status as reflected by CRP levels in blood. Mathematical models that aim to predict the balance that exists between immune activation and regulation within accessible blood samples, will therefore benefit from additionally taking into account the following variables: i) the effect of diurnal variation ii) the tumour growth rate iii) the heterogeneity present within both regulatory and effector T cell populations.

In conclusion, in our sample of patients with gynaecological malignancies, CRP concentrations do not oscillate in a consistent predictable manner, and do not correlate either positively or negatively with conventional T_reg or T_eff subsets. Therefore, there is no evidence to suggest that CRP can reliably be used across cancer patients as a surrogate, time-sensitive and most importantly, predictive marker, to reflect circulating effector or regulatory T cell frequencies, as previously suggested [13]. Time based therapy founded on modelling a consistent cyclical pattern of inflammation using serum CRP concentration as a predictive marker of regulatory T cell expansion may not be possible. However, we cannot exclude that further investigating the kinetics of inflammation in cancer patients, perhaps by taking more frequent blood samples or else by taking into consideration multiple inflammatory and immune-regulatory parameters, the progressive nature of immune suppression, as well as the heterogeneity of effector and regulatory T cell populations could all help in modelling more complex, and potentially predictive, equations.