INTRODUCTION

In sports, injury prevention is a primary concern for teams and clubs’ technical committees and health departments, not only because of the damage caused by the athlete’s distance itself but also because of the high costs arising from the underlying treatments (^{Neves et al., 2016}; ^{Viegas et al., 2020}). Considering this point of view, especially in high-performance sports activities, there is great interest in technical committees in facilitating and cheapening the process to identify possible future injuries of an athlete through the technology available today (^{Santos Bunn, 2020}).

Thereunto, there are several direct or indirect indicators of the possibility of injury that can be used to monitor the health status and physical strain of athletes, such as measuring enzymes present in the bloodstream — lactate dehydrogenase (LDH), troponin I, myoglobin, and creatine kinase (CK-MM) — verification of maximum voluntary action, ultrasonography, tomography, surface electromyography, magnetic resonance, muscle biopsy, thermography, observation of subjective pain perception scale, also represented by delayed onset muscle soreness (DOMS), among others (^{Andrade Fernandes et al., 2017a}).

The CK, in its muscular form, is an objective measure. However, it can only be accessed through the blood, not allowing for a precise locate the injured structure. In this sense, the combination of indicators that will enable objective and subjective measurements have been recommended, and several studies have pointed out the use of thermography for this purpose (^{Andrade Fernandes et al., 2017b}; ^{Júnior et al., 2021}; ^{Neves et al., 2015}; ^{Neves et al., 2016}; ^{Priego-Quesada et al., 2020}).

In case of suspicion of musculoskeletal injury, imaging tests (ultrasound, tomography, surface electromyography, magnetic resonance imaging, among others) have been the main choice for confirming the diagnosis. However, they are costly methods to be maintained periodically, in addition to not being portable and emitting excessive ionising radiation, which is harmful to health. Therefore, a solution to promote the optimisation of this process is the use of thermography. It is a tool that obtains infrared images through cameras capable of measuring the surface temperature of the body’s skin, indicating an accelerated metabolic activity in some areas of the body, which may show the location of sites with a high chance of developing lesions (^{Andrade Fernandes et al., 2017a}). The advantages are vast compared to other imaging techniques: safety, non-contact, non-invasive, painless, no ionising radiation, no side effects, no contraindications, and efficacy (^{Viegas et al., 2020}). Researchers have, therefore, been successful in the combined use of physiological indicators to prevent injuries. If, on the one hand, high serum CK levels indicate the possibility of injury in any body area, thermography allows locating this area so that, consequently, a prevention protocol can be adopted (^{Andrade Fernandes et al., 2017b}; ^{Júnior et al., 2021}; ^{Kennedy et al., 2017}; ^{Neves et al., 2015}; ^{Neves et al., 2016}; ^{Priego-Quesada et al., 2020}).

However, the literature doesn’t clarify whether the variation shown in skin temperature is associated with indirect blood markers of muscle damage, as some studies find an association (^{Andrade Fernandes et al., 2017a}; ^{Andrade Fernandes et al., 2017b}; ^{Júnior et al., 2021}) and others do not (^{Bandeira et al., 2012}; ^{Kennedy et al., 2017}). Knowing the magnitude of the associations between CK and thermography would support a more scientific, practical, and efficient application to prevent injuries in physically active subjects. In this sense, it can be questioned whether an eventual direct replacement of the CK assessment (invasive method) by the thermal image assessment (non-invasive method) would be possible? Or whether both should be used in a complementary way?

Thus, this study aimed to investigate the association between thermal variation identified by thermography and muscle CK values in healthy subjects from a systematic review with meta-analysis. A null hypothesis (h0) assumed that there was no association between thermal variation identified by thermography and muscle CK values.

METHODS

RESULTS

The PRISMA flow diagram of the included studies is depicted in Figure 1 below. In Figure 1, it can be seen that 482 files were identified in the databases, and another one was obtained by manual search. After reading the titles and abstracts, 386 duplicates were excluded. Another 65 records marked as ineligible by automation tools (search filters) were removed from areas of knowledge and articles older than the year 2000. In addition, of the remaining 31 articles, another 15 were removed for not specifically addressing the topic sought and another four for not having a full version in English, leaving 13 reports evaluated as eligible to be included in the scope of study of the systematic review. Of the 13, only five had a correlation coefficient “r” referring to the topic and, therefore, were included in the meta-analysis.

Figure 1. Flow diagram of the studies included in this systematic review with meta-analysis regarding the association between CK and thermography. 

The characteristics of the 13 studies included in the systematic review are presented in Table 2, which details the characteristics of the samples, the intervention methods used, the regions of interest (ROI) analyzed by thermal imaging, the results obtained, and the conclusions.

From the related samples, all studies selected healthy, non-injury professional athletes, with general ages ranging from 15 to 47 years, and from sports such as soccer, rugby, cross-country skiers, and endurance swimmers, except four studies that selected amateur runners (^{Pérez-Guarner et al., 2019}), untrained men (^{Da Silva et al., 2018}), and just healthy individuals (^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}). The selected ROI, in turn, were in short anterior and posterior chains of the lower limbs, except for three studies that included anterior and posterior views of the whole body (^{Drzazga et al., 2018}; ^{Pérez-Guarner et al., 2019}; ^{WHO, 2014}) and two more that selected only the biceps (^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}) as the region of interest.

Regarding the intervention methods, all studies performed infrared thermal imaging and blood collection, obtaining, respectively, the surface temperature of the skin in the ROI and the serum levels of CK and other biomarkers. Some studies also measured other variables such as countermovement jumping (^{Júnior et al., 2021}; ^{Pérez-Guarner et al., 2019}), perception of recovery (^{Carvalho et al., 2021}), perception of fatigue (^{Barros et al., 2020}; ^{Carvalho et al., 2021}; ^{Pérez-Guarner et al., 2019}), perception of pain (^{Barros et al., 2020}; ^{Carvalho et al., 2021}; ^{Da Silva et al., 2018}; ^{Pérez-Guarner et al., 2019}; ^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}), functional markers (heart rate, the volume of maximum oxygen, etc.) (^{Drzazga et al., 2018}) and peak torque (^{Barros et al., 2020}). In addition, each author performed the assessments in different time frames. Regarding stimuli (sports matches, aerobic exercises, counter-resistance exercises, etc.), the authors performed the assessments as follows: 48 h before (^{Pérez-Guarner et al., 2019}), 24 h before (^{Andrade Fernandes et al., 2017a}; ^{Andrade Fernandes et al., 2017b}; ^{Pérez-Guarner et al., 2019}), immediately before (^{Bandeira et al., 2012}; ^{Barros et al., 2020}; ^{Da Silva et al., 2018}; ^{Drzazga et al., 2018}; ^{Júnior et al., 2021}; ^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}), immediately after (^{Bandeira et al., 2012}; ^{Da Silva et al., 2018}; ^{Júnior et al., 2021}; ^{Neves et al., 2016}), 1 h after (^{Drzazga et al., 2018}), 3 h after (^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}), 24 h after (^{Andrade Fernandes et al., 2017a}; ^{Andrade Fernandes et al., 2017b}; ^{Bandeira et al., 2012}; ^{Carvalho et al., 2021}; ^{Matta et al., 2019}; ^{Pérez-Guarner et al., 2019}; ^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}), 48 h after (^{Andrade Fernandes et al., 2017a}; ^{Andrade Fernandes et al., 2017b}; ^{Bandeira et al., 2012}; ^{Barros et al., 2020}; ^{Da Silva et al., 2018}; ^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}) and 72 h after (^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}).

Table 3 below shows the quality assessment of studies carried out with the QUADAS-2 tool (^{Whiting et al., 2006}) by analysing the risk of bias and concerns about the applicability and external validity of the studies, classifying them as low risk (↓), non-risk evident (?) and high risk (↑).

Considering the 13 studies reviewed, there is a general balance in obtaining results of CK increase along with skin temperature versus studies that didn’t observe any association between the measures. In other words, five studies (^{Andrade Fernandes et al., 2017a}; ^{Andrade Fernandes et al., 2017b}; ^{Júnior et al., 2021}; ^{Matta et al., 2019}; ^{Pérez-Guarner et al., 2019}) found some association between the variables and eight studies (^{Andrade Fernandes et al., 2017b}; ^{Bandeira et al., 2012}; ^{Carvalho et al., 2021}; ^{Da Silva et al., 2018}; ^{Matta et al., 2019}; ^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}) didn’t find any type of association.

The result of the meta-analysis is shown in Figure 2 below. Each line of the graph, generated in the Stats Direct 3.0 tool, indicates the Pearson or Spearman correlation coefficient in each region of interest analyzed by thermal imaging in certain temporal sections evaluated by each author, with their respective p values inserted. In addition, the I² statistic = 0% (95%CI 0% – 25.8%) was observed, indicating that there isn’t evidence of heterogeneity between studies. Thus, according to Figure 2, the correlation coefficient of the combined studies presented r= –0.05 (95%CI –0.11 – 0), with p-value= 0.0568, which indicates a result not significant for the association between CK and thermography.

Figure 2. Hedge-Olkins Fixed Effects method meta-analysis plot: combined association between CK and thermography. Lowercase letters indicate different regions of interest (ROIs) evaluated in the same study. 

DISCUSSION

The present study evaluated the statistical association between serum CK levels and skin temperature through a systematic review with meta-analysis. A total of 483 studies were retrieved from the databases, and 13 were selected for qualitative analysis, five of which were selected for quantitative analysis (Figure 1).

Most of the authors (^{Andrade Fernandes et al., 2017b}; ^{Bandeira et al., 2012}; ^{Da Silva et al., 2018}; ^{Matta et al., 2019}) who claimed not to find associations didn’t present the corresponding statistical data. Furthermore, there was one author (^{Pérez-Guarner et al., 2019}) who found an association is only one of the chosen ROIs and presented this only result, failing to present the non-association results obtained from the other ROIs. Some authors (^{Andrade Fernandes et al., 2017b}; ^{Barros et al., 2020}; ^{Drzazga et al., 2018}; ^{Júnior et al., 2021}; ^{Matta et al., 2019}; ^{Pérez-Guarner et al., 2019}; ^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}) presented general results, without providing the association in each chosen ROI.

Comparing the reviewed studies, the fact of obtaining a general balance between positive and negative results for the association between variables observed by the systematic review reinforces that there isn’t consensus in the literature and that, consequently, there isn’t a way to categorically state that there is any relationship between the variables. Da Silva et al. (²⁰¹⁸), for example, observed that there was no association between the two variables after a muscle injury induction protocol; it can be said that in a real situation, with the analysis of training diaries and comparing with laboratory situations, it wasn’t possible to see an association between muscle damage biomarkers with increased skin temperature. However, it is worth noting that a good part of the studies, such as the one by Andrade Fernandes et al. (^2017b) for example, found significant associations with the justification that strenuous exercise does generate micro muscle injuries that cause muscle CK to leak into the bloodstream, as well as lead to inflammatory processes that increase local temperature. However, studies recommend caution when interpreting the results, as well as reinforcing the idea of using thermography as a joint diagnostic support method, as Bandeira et al. (²⁰¹⁴) recommends for example.

Some studies performed collections 48 h before (^{Pérez-Guarner et al., 2019}) and 24 h before (^{Andrade Fernandes et al., 2017b}; ^{Neves et al., 2015}; ^{Pérez-Guarner et al., 2019}), as well as immediately before (^{Andrade Fernandes et al., 2017b}; ^{Bandeira et al., 2012}; ^{Barros et al., 2020}; ^{Da Silva et al., 2018}; ^{Drzazga et al., 2018}; ^{Júnior et al., 2021}; ^{Matta et al., 2019}; ^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}) and after (^{Bandeira et al., 2012}; ^{Barros et al., 2020}; ^{Da Silva et al., 2018}; ^{Júnior et al., 2021}) and also 1 h after (^{Drzazga et al., 2018}), 3 h after (^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}), 24 h after (^{Andrade Fernandes et al., 2017a}; ^{Andrade Fernandes et al., 2017b}; ^{Bandeira et al., 2012}; ^{Carvalho et al., 2021}; ^{Matta et al., 2019}; ^{Pérez-Guarner et al., 2019}; ^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}), 48 h after (^{Andrade Fernandes et al., 2017a}; ^{Andrade Fernandes et al., 2017b}; ^{Bandeira et al., 2012}; ^{Barros et al., 2020}; ^{Da Silva et al., 2018}; ^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}) and up to 72 h after (^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}), with each author choosing a period they deemed convenient for their intervention, causing divergences in the general comparison of studies. In addition, some authors (^{Bandeira et al., 2012}; ^{Barros et al., 2020}; ^{Drzazga et al., 2018}; ^{Júnior et al., 2021}; ^{Petrofsky et al., 2011}; ^{Petrofsky et al., 2012}), despite having performed collections in different periods, presented statistical results of general association, while others presented results by period specifically (^{Andrade Fernandes et al., 2017b}).

This difference in the time cut used by each author (1 h, 3 h, 24 h, 48 h, etc. after intervention) may impact the interpretation of the results since CK has its peak incidence in the bloodstream within 24 h after the stimulus, with cases in which it can increase or decrease within 72 hours (^{Bandeira et al., 2012}). Thus, it is considered that this non-linearity of the biomarker and the fact that some blood collections were performed outside this time range after the stimulus may have affected the association between the observed CK level and skin temperature. However, heterogeneity wasn’t observed among the included studies.

The results showed that thermographic evaluation isn’t a recommended substitute for tests to detect the indirect biomarker CK in blood since no significant association was observed in the evaluation by meta-analysis (r= –0.05; p= 0.0568; 95%CI –0.11 – 0). In addition to the divergence in the design of the studies analyzed, this result can also be explained based on the physiology of each of these indicators. CK is a systemic biochemical indicator present in the bloodstream, and its values can be altered by damage to inner or outer muscles. On the other hand, although body temperature is also systemic, surface skin temperature (measured by thermography) is a local indicator (^{Fernández-Cuevas et al., 2015}), reinforcing the lack of association.

In addition, skin temperature can be influenced at times due to inflammation or infection, fractures, oedema, etc. (^{Fernández-Cuevas et al., 2015}). CK values, on the other hand, aren’t necessarily modified by these types of alterations, except when they result from a muscle injury (^{Bandeira et al., 2012}). Another factor to be considered is that each individual may present different levels of this biomarker, and some individuals may reach this CK peak in the bloodstream within 72 hours (^{Bandeira et al., 2012}).

Bandeira et al. (²⁰¹²) already stated that thermography associated with the use of CK could be used as a complementary method for locating muscle injuries in athletes, suggesting, however, opting for visual inspection of thermal images. The results found in this meta-analysis reinforce the practical applicability of thermography, like the fact that no significant association was found between serum CK levels, as well as the association between thermographic assessment and the risk of injury (^{Santos Bunn et al., 2020}), which reinforces the idea of using thermal imaging in sports as a complementary method to more invasive exams for the assessment of athletes and, mainly, in the diagnosis and prevention of injuries, because when there is an association between two methods, it isn’t necessary to use both, as from one of them it would be possible to estimate the other.

Thus, the main limitations of the evidence included in the review are related to the limited provision of statistical data, as well as the lack of magnitude and quantity of results. When comparing serum CK levels and thermography, there is a lack of presentation of correlation coefficients.

Limitations of this study can be mentioned: the scarcity of studies specifically looking for the association between serum CK levels and thermography, the diversity of different study designs, and, as mentioned in the previous paragraph, the lack of objective records of the relationship between serum CK levels and skin temperature.

To provide a perspective for future studies, we suggest the scientific community carry out further research seeking the association between serum CK levels and thermography in sports that have a low possibility of trauma due to contact between athletes since this can represent an intervening variable in the study impacting the interpretation of infrared images. Another suggestion is to search for this association in sports activities, whose primary motor muscles are superficial, such as cycling and hard running, making it easier and more accurate to identify, through thermal images, possible muscle damage that will also lead to an increase in CK.

CONCLUSION

Considering the parameters related to the use of thermography in this systematic review with meta-analysis, it can be concluded that there is no association between serum CK levels and thermography results. In clinical practice, it is recommended to use infrared thermal images in complement to the assessment of serum CK levels to assist in the determination of differential diagnoses of musculoskeletal injuries.