Natural killer cell levels in adults living with type 2 diabetes: a systematic review and meta-analysis of clinical studies

Background Chronic immune activation and hyperglycaemia are a hallmark of type 2 diabetes mellitus (T2D) while natural killer (NK) cells are involved in the pathogenesis of T2D. Dysregulated NK cell responses are associated with an increased risk of cardiovascular disease in patients living with T2D. Objective To provide a comprehensive and systematic evidence-based estimate on the levels of NK cells in patients living with T2D. Results This systematic review and meta-analysis included 13 studies reporting on 491 adult patients with T2D and 1064 nondiabetic controls. The pooled effect estimates showed increased levels of NK cells in adult patients with T2D compared to controls (MD: 0.03 [− 3.20, 3.26], I2 = 97%, p < 0.00001). Conclusion Overall, the evidence presented in this systematic review shows that the changes in NK cells in patients living with T2D are still unclear and further studies are needed.


Background
Non-communicable disease such as type 2 diabetes (T2D) are amongst the leading causes of mortality and continue to burden the healthcare system of both developing and developed countries [1]. Globally, more than 425 million adults are living with diabetes, with T2D accounting for approximately 90% of the cases [2]. In developing countries such as those in sub-Saharan Africa, the prevalence of T2D is expected to increase by more than two-fold in the next 30 years and this coincides with an increased risk of cardiovascular disease (CVD) [3]. Obesity has been identified as a major risk factor for T2D [4].
Notably, over 90% of patients with T2D are obese [5] and these individuals present with ectopic lipid accumulation that is associated with adipose tissue (AT) dysfunction [6]. Although mechanisms involved in AT dysfunction in T2D are complex, enhanced infiltration of leukocytes into the AT promotes low-grade chronic inflammation which is mediated by T-helper (Th) 1 and Th17 responses [7]. Interestingly, our group and others have provided cumulative findings showing that persistent Th1 and Th17 cytokine levels which include tumor necrosis factor-α (TNFα), interleukin-1 (IL-1), IL-6 and IL-17 exacerbates insulin resistance which may ultimately lead to cardiovascular complications [8][9][10][11][12]. Furthermore, natural killer (NK) cells are constituents of the AT resident immune cell population which orchestrate a localised inflammatory response [12,13]. A dysregulation of these immune cells promotes insulin resistance [13,14].
There has been a growing interest in the role of NK in T2D, and in the development of T2D associated complications. In fact, a direct association between elevated NK cell levels and an increased risk of CVD has been reported [15]. Contradictory findings on the role of NK cells in exacerbating inflammation and insulin resistance in T2D exist [16,17]. Thus, the role of NK cells in modulating T2D and its associated complications remains elusive.
There is currently no meta-analysis available aimed at providing an estimate of the levels of NK cells in patients living with T2D, who may also at risk of developing CVD. Therefore, this systemic review and metaanalysis aimed at providing a comprehensive synthesis of available literature elucidating the role of NK cell function in T2D. Since cardiovascular complications remain amongst the leading causes of death in patients with T2D [17] it is crucial to establish the relationship between NK cells and the development of T2D.

Study selection
We identified a total of 40 citations (38 MEDLINE and 2 from other sources). A total of 14 citations were excluded after screening the title and the abstracts (Fig. 1). The remaining 26 studies were assessed for full-text eligibility. In all, 7 reviews and 6 studies were excluded as these were irrelevant (Fig. 1). We reported the findings of this systematic review and meta-analysis following the Meta-analysis Of Observational studies in epidemiology (MOOSE) guidelines [18]. In total, 13 studies fulfilled the study selection criteria and were included in the qualitative analysis [14-22, 24, 25, 27, 28, 35]. While only six studies were included in the meta-analysis [14,15,[19][20][21]35].

Primary outcomes Levels of natural killer cells in patients living with T2D
A total of six studies [16,20,[22][23][24][25] reported no difference in the frequencies of circulating NK cells between patients with T2D and controls. Whilst others reported contradictory findings of decreased [25,35] and increased [14] NK cells levels in patients with T2D. Notably, NK cells in patients with T2D expressed higher levels of the early activation marker CD69 [35] despite the reported lower frequency in T2D [15,35]. To further ascertain the robustness of our reported effect estimate of elevated levels on NK cells in T2D, we performed a sensitivity analysis based on participants' smoking status which was previously reported to influence the levels of NK cells [29]. As a result, we omitted a study by Olson and colleagues which included T2D patients that were smokers and our results remained consistent and the level of statistical heterogeneity reduced to 0% (MD: 4.81 [4.70, 4.92]; I 2 = 0%) (Table 2S). Notably, the potential sources of bias in the included studies were mainly due to differences in the external validity of the studies (Table 3S). In contrast, the selection and reporting bias ratings were comparable between the included studies (Table 3S). In the external validity sub-scale, studies that had a low risk of bias followed procedures to ensure that the included participants were representative of the population [14,19,20]. Whereas none of the studies that were rated as high risk reported on steps performed to validate the representativeness of the source population [15,21,35]. These studies obtained an overall score of 0 for the external validity domain (Table 3S) (Table 2).

Discussion
Delineating the pathological mechanisms involved in the development of T2D and its related complications remains crucial in the discovery of novel biomarkers of early disease, including establishing effective therapeutic strategies. There has been a growing interest in understanding the role of immunological mechanisms in the prognosis of inflammatory conditions. NK are the major innate lymphocyte subsets that are involved in the  [30,31]. The aim of this systemic review and meta-analysis was to provide a comprehensive synthesis of available studies reporting on NK cells in adult patients living with T2D.
The cumulative evidence presented in this study shows that the levels of NK cells are slightly elevated in patients with T2D [15,21,36]. However, there seems to be distinct difference in the regulation of NK cells in T2D [16,17]. In particular, some studies reported on the comparable levels of NK cells between patients with T2D and nondiabetic controls [15,16,20,23,24]. However, peripheral blood NK cells expressed higher levels of CD69, an early activation marker that is essential for the modulation of inflammatory response [35]. In addition to CD69, NKG2D is another surface receptor, mainly identified in NK cells that is elevated in patients with T2D patients [15], although others reported lower levels of NKG2D in patients with T2D [16,19]. The regulation of NKG2D in patients with T2D remains unclear. However, the stimulation of NK cells through NKG2D or NKp46, its activation receptor, may exacerbate the production of inflammatory cytokines [16]. Moreover, studies have reported on elevated levels of the inhibitory markers NKB1, CD158b, and KIR3DL-1 in patients with T2D when compared to controls [15,19,25,26,35].
These are inhibitory markers on the surface NK cells that regulate cytotoxic activity these cells. Furthermore, an imbalance between activation and inhibitory receptors expressed on NK cells may compromise immune tolerance resulting in β-cell destruction [15]. In fact, pancreatic β-cells are destroyed as a consequence of an inflammatory process initiated by lymphocytes of the immune system [40]. Although the evidence presented in this systematic review and meta-analysis, suggest that NK cells are activated in patients with T2D, only a few studies reported on effect measures associated with the pathogenesis of T2D. It is well-established that patients with T2D present with dysregulated immune responses and dyslipidaemia, as a consequence of altered lipid metabolism which is associated with an increased risk of developing CVD [41].
The evidence presented in this systematic review and meta-analysis was limited by the unavailability of studylevel data of effect measures in some of the included studies, which affected pre-specified quantitative analysis described. Moreover, majority of the included were cross-sectional studies with a high risk of bias, particularly in the internal and external validity domains. Therefore, caution should be taken in the interpretation of these findings and extrapolation of these findings in different populations.

Conclusion
Overall, the evidence presented in this review suggests that the changes in NK cell counts in patients living with T2D remain unclear and further studies are needed.

General guidelines applied in the current systematic review
This systematic review and meta-analysis was reported according to the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines [18]. The review protocol was also registered on PROSPERO (CRD42018106159) and has been published [36]. We included randomised control

Search strategy and study selection
A comprehensive search strategy was developed in consultation with a subject librarian (Table 1S). The search was conducted to identify and retrieve studies reporting on relevant findings that would address the following research questions: Research question: Are NK cells elevated in patients with T2D?
The search strategy comprised of medical subheadings (MeSH) and corresponding keywords which included; type 2 diabetes mellitus, natural killer cells, and cardiovascular disease. The search strategy was used to search the MEDLINE, Cochrane Library and Embase electronic databases using the PUBMED and google scholar search engines. The search was restricted to available full-text human studies published from inception up until 13 November 2019. We also searched the OpenGrey (System for Information on Grey Literature in Europe) (www. opengrey.eu) although these were not included, the bibliographies were screened for relevant citations. In addition, the bibliography of included studies was also screened for additional relevant studies. We used the Mendeley referencing manager (V1.19.10) to manage the included studies and to remove duplicates.

Primary outcomes
The levels of peripheral blood NK cells were the primary outcome of this systematic review and meta-analysis.

Risk of bias and quality of the studies
The potential risk of bias in the included study was independently assessed by two reviewers (VM and TMN) using the modified Downs and Black checklist [37] which has four domains namely; reporting bias (10 items), external validity (3 items), internal validity (6 items) and selection bias (7 items). In instances of disagreements, a third reviewer (BBN) was consulted for arbitration. The studies were rated as; excellent if the score was (24-28 points), good if (19-23 points), fair if (13-18 points) and poor if (≤12 points). The levels of inter-rater agreement were assessed using the Cohen's kappa [37] and the funnel plots were used to assess publication bias.

Statistical analysis
The meta-analysis of pooled data was performed using Review Manager V.5 (Cochrane Collaboration, Oxford, UK). The level of statistical heterogeneity among the included studies was assessed using Higgin's I 2 index statistic. An I 2 value of > 50% was considered as substantial heterogeneity [38]. To explore the sources of heterogeneity within the included studies, a subgroup analysis was performed based on the effect measures reported. Random-effects model was used in the meta-analysis due to substantial levels of heterogeneity amongst the included studies. Moreover, effect estimates were reported as the mean difference (MD) or the standardised mean difference (SMD), and the 95% confidence. The Cohen's kappa was used to assess the inter-rater reliability and a p-value of less than 0.05 was statistically significant.
Innovation Platform, Medical Research Council, Tygerberg, South Africa.