All authors helped define the protocol for this systematic review, which was developed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols (PRISMA-P) (16). In addition, this protocol was registered with the number CRD42021260764 (https://www.crd.york.ac.uk/PROSPERO/view/CRD42021260764) in the Prospective International Registry of Systematic Reviews (PROSPERO) (17).

To prepare and structure this review, the focused question was formulated using the PICO format (population, intervention, comparison and outcomes) as detailed below: •

Population: People with oncological and infectious diseases

What are the salivary biomarkers available as a prognosis for oncological and infectious diseases?

For the present systematic review, a bibliographic search was carried out in 8 electronic databases (Pubmed, Cochrane Library, Scopus, EMBASE, Web of Science (WoS), Scielo, Sciencedirect and Google Scholar) until July 2025; combining keywords and subject titles according to the thesaurus of each database: “saliv*”, “cytokine”, “biomaker”, “interleukin”, “neoplasms”, “HIV”, “Epstein -Barr virus”, “EBV”, “tuberculosis” and “clinical trial”. The search strategies of each of the databases are found in Table 1.

Additional relevant literature was also included through manual review of the reference lists of the finally selected articles.

The search of electronic databases was performed independently by two reviewers (HA and SP), and the inclusion of studies was determined based on predefined criteria: randomised clinical trials (RCTs) published in English, with no restriction on the date of publication, and reporting on the use of salivary biomarkers for prognostic purposes in oncological and/or infectious diseases.

Data extraction from each eligible study was performed using a structured template that included key variables such as year of publication, authorship, country of origin, study title, sample size, gender distribution, associated systemic condition, mean age, group allocation, salivary biomarkers assessed, analytical methods used, presence of additional biomarkers, follow-up duration, and conclusion. Two independent reviewers (FCO and HV) performed the data extraction, and any discrepancies were resolved by consensus with a third reviewer (FCZ).

Using the Cochrane Risk of Bias 2.0 tool, two calibrated reviewers (RA and JM) separately assessed the risk of bias (RoB) of the included studies (18), and a high level of inter-reviewer agreement was observed (k = 0.98). Randomisation procedure, deviations from planned interventions, lack of outcome data, outcome assessment, and selection of reported outcomes are the five categories in which this instrument evaluates randomised clinical trials. Studies are classified as low risk of bias, moderate concerns or high risk of bias according to these factors.

A qualitative evaluation of the included studies’ findings was the sole focus of the synthesis. Furthermore, GRADE analysis was used to evaluate the evidence's certainty using the GRADEPro GDT guideline development tool, which was created by McMaster University and Evidence Prime Inc. in Canada.

A comprehensive manual and electronic search initially identified 372 records, of which 47 duplicates were removed (Figure 1). After reviewing the titles and abstracts, 26 full-text articles were considered potentially eligible and retrieved for detailed assessment. Of these, 10 did not meet the inclusion criteria and were excluded. Consequently, 16 randomised controlled trials were considered eligible and included in the qualitative synthesis. The specific reasons for exclusion are indicated in Table 2.

Overall, 16 RCTs (29–44) with publication years ranging from 2008 to 2024 were included. The countries where the studies were conducted were: United States (31–35, 37, 39, 43, 44), South Africa (36), Brazil (29, 30, 38, 39), United Kingdom (41) and Venezuela (41) (Table 3).

The number of patients in all studies ranged from 14 to 322. Systematic diseases or conditions in all studies were: cancer (29–33, 35, 37–42, 44), HIV (34, 43) and tuberculosis (46); having as treatments drugs (33, 34, 36, 42), relaxation exercises and stress reduction (31, 32, 35, 37, 40, 41, 43, 44) and laser (29, 30, 38, 39). The mean age was reported in 14 studies (29, 31–37, 39–41, 43, 44) which ranged between 34.49 and 67 years. Follow-up time in all studies ranged from 4 h to 6 months. The most used salivary biomarkers were cortisol and interleukins (Table 3).

Ten studies (29, 30, 32, 34–36, 38–41) were at low risk of bias, and 6 studies (31, 33, 37, 42–44) were at high risk of bias (Figure 2).

The most frequent sources of bias were related to blinding of outcome assessors, incomplete outcome data, and lack of pre-specified protocols. Studies using biomarker assays without standardized saliva collection protocols showed increased detection bias. Among the six studies rated with high risk of bias, most reported positive findings, raising the possibility of publication or reporting bias.

The GRADE assessment indicated low overall certainty of the evidence, primarily due to serious concerns related to risk of bias and inconsistency. Several studies lacked robust randomization methods or blinded outcome assessments. Inconsistencies in assay methods and outcome definitions contributed to heterogeneity in findings. Although indirectness and imprecision were not serious issues, the lack of external validation and variability in sample sizes limited the confidence in the clinical applicability of the findings (see Table 4).

The 16 included randomized controlled trials explored the use of various salivary biomarkers for monitoring oncological and infectious diseases, predominantly focusing on cytokines such as IL-6, IL-1β, IL-8, TNF-α, cortisol, and CRP. These biomarkers were evaluated in contexts ranging from cancer-related fatigue and stress responses to infectious disease progression in HIV and tuberculosis patients.

Among the studies, interleukins (particularly IL-6 and IL-1β) were frequently used to reflect proinflammatory activity. Cortisol levels were studied in relation to psychoneuroimmunological responses to disease and stress interventions, such as relaxation therapy. While methodological approaches varied, common findings included the potential diagnostic and prognostic value of these biomarkers, despite challenges like salivary variability, sample handling, and lack of standardization across assays.

To aid interpretation, Table 5 presents a comprehensive overview of each major biomarker's biological relevance, clinical utility, methodological limitations, and proposed directions for future research. This synthesis aims to guide researchers and clinicians in prioritizing salivary biomarker candidates and designing standardized protocols for future applications.

This systematic review included 16 randomized controlled trials evaluating the prognostic utility of salivary biomarkers in oncological and infectious diseases. The most commonly reported biomarkers were interleukins (especially IL-1β, IL-6, IL-8, and IL-10) and cortisol, which were frequently elevated in patients with systemic inflammatory conditions and reduced following therapeutic interventions. These findings highlight the growing potential of saliva as a non-invasive medium for monitoring disease progression and treatment response in diverse clinical scenarios (45, 46).

The clinical relevance of salivary biomarkers lies in their ability to reflect not only local but also systemic pathophysiological changes, offering a promising adjunct to traditional diagnostic and monitoring methods. For example, IL-6 is known to exhibit both pro- and anti-inflammatory properties and is actively involved in B and T cell differentiation, while IL-8 contributes to neutrophil recruitment and activation in acute inflammation (35). Cortisol, a key stress hormone, was found to be elevated in patients with oncological and infectious conditions and decreased post-intervention in several trials, reflecting its utility as a stress-responsive prognostic marker (47). While IL-6 has dual functions depending on the cellular context and phase of immune response, the presence of elevated levels in saliva generally reflects its pro-inflammatory role during active disease states (48). This duality does not contradict its elevation in pathological conditions, as it often predominates in a pro-inflammatory profile during acute and chronic inflammation.

The interleukins most frequently reported across the included studies were IL-1β, IL-6, IL-8, TNF-α, and IL-10, all of which are central mediators of the inflammatory response. These cytokines were found to fluctuate in response to treatments such as photobiomodulation therapy (PBMT), pharmacological agents, and stress-reduction techniques. For example, IL-1β and TNF-α are typically elevated during the acute phase of inflammation, and their salivary levels were shown to decrease post-treatment in several cancer-related interventions (29–30, 38–39).

IL-6, a multifunctional cytokine with dual pro- and anti-inflammatory roles, was also widely assessed. Its regulation in saliva following oncological treatments or behavioral interventions suggests potential for monitoring immune recovery and systemic burden (32, 35, 37). Similarly, IL-8 plays a causal role in neutrophil activation and was found to correlate with disease activity and treatment response in both cancer and infectious diseases (33, 36, 42).

The inclusion of salivary cortisol in several studies highlights its role as a reliable biomarker for stress-related physiological changes. Its reduction after interventions such as cognitive behavioral stress management (CBSM), mindfulness-based stress reduction (MBSR), and yoga practices reflects its dynamic nature and prognostic significance (31–32, 35, 37, 40–41, 44).

Notably, salivary cortisol may offer real-time monitoring of hypothalamic–pituitary–adrenal (HPA) axis activity, and its immediate elevation following stress stimuli and subsequent reduction post-intervention make it a valuable marker for evaluating the systemic impact of disease and the effectiveness of therapy (49).

Several studies listed in Table 3 included nitrite as a salivary biomarker, primarily in relation to inflammatory modulation. Beyond its role as an indicator, nitrite may hold promise as a prognostic bioactive factor due to its involvement in nitric oxide metabolism, oxidative stress, and immune response regulation. Nitrite has been linked to tumor progression and therapeutic responsiveness, particularly in head and neck cancers, making it a candidate for future prognostic profiling in oncological contexts (50).

Although both oncological and infectious diseases triggered elevated salivary cytokines and cortisol levels, the patterns of biomarker expression and their response to treatment varied depending on the underlying condition. This elevation does not imply a linear or static increase; rather, cytokine levels, particularly those like IL-6, may vary dynamically depending on disease stage, immune response, or treatment effect.

In oncological diseases, such as head and neck cancer, breast cancer, or leukemia, cytokine profiles typically reflected chronic inflammation and immune dysregulation. Interventions including PBMT, chemotherapy, or relaxation programs were associated with moderate to significant reductions in salivary IL-1β, IL-6, IL-8, and TNF-α (29–32, 35–39, 42). These reductions often coincide with clinical improvements, such as reduced oral mucositis or decreased self-reported stress. This supports the idea that salivary biomarkers may serve not only as biological indicators of inflammation but also as clinical correlates of treatment efficacy in cancer patients (2).

In contrast, infectious diseases such as HIV and tuberculosis showed a more variable pattern. For example, Rubin et al. (34) and Dudgeon et al. (43) reported acute increases in cortisol and interleukins shortly after pharmacological or physical interventions, which may reflect the immediate immune activation seen in infectious processes or early treatment responses. Similarly, Shenje et al. (36) observed complex shifts in multiple interleukins, such as IL-6 and IL-10, following corticosteroid therapy in tuberculosis patients. These findings suggest that the time course and interpretation of salivary biomarkers in infectious diseases may require more nuanced approaches and time-point–specific sampling protocols (51).

While both disease categories activate systemic immune responses, oncological patients often present with sustained elevation of biomarkers due to tumor-induced inflammation, whereas in infectious cases, fluctuations may be more rapid and transient, related to microbial load or treatment onset. These fluctuations reflect the evolving nature of immune response, where cytokine levels may rise during disease exacerbation and decline with effective treatment or resolution of infection. In oncological conditions, this modulation may follow therapeutic cycles, whereas in infections, fluctuations often correspond to microbial burden or host immune activation. Therefore, clinicians must consider disease-specific biomarker dynamics when using saliva for prognostic purposes.

The use of salivary biomarkers such as interleukins and cortisol offers promising clinical applications in the monitoring and follow-up of patients with oncological and infectious diseases. Due to its non-invasive, painless, and low-risk nature, saliva collection is particularly suitable for patients undergoing immunosuppressive treatments, those with reduced venous access, or in pediatric and geriatric populations (48).

In clinical practice, these biomarkers could help in: •

Assessing therapeutic response, particularly in stress-reducing interventions and anti-inflammatory treatments.

The findings of this review suggest that salivary biomarkers—particularly cortisol and inflammatory interleukins—may serve as useful adjuncts in the prognostic evaluation and therapeutic monitoring of patients with oncological and infectious diseases. Their non-invasive nature allows for repeated monitoring of systemic inflammation and stress-related responses, especially in vulnerable populations such as cancer patients and individuals with immunodeficiency. These biomarkers offer the potential for dynamic assessment of therapeutic effectiveness, particularly in interventions targeting inflammatory or psychological components, including corticosteroid therapy, cognitive-behavioral interventions, or photobiomodulation. Moreover, their use can improve patient compliance and comfort by avoiding the discomfort and risk associated with blood draws, especially when frequent sampling is required. In resource-limited, outpatient, or remote care settings, salivary biomarker testing may offer a practical and accessible alternative to traditional laboratory methods. Integration of salivary diagnostics into oncology clinics, infectious disease units, and primary care settings could facilitate early detection of relapse, identification of treatment-related complications, and personalized therapeutic adjustments. However, to enable their successful clinical adoption, further efforts are needed to ensure proper validation, cost-effectiveness analyses, and training for healthcare professionals in the interpretation and use salivary biomarker assays (6).

To enhance interpretability and guide future translational research, a comparative summary of the key salivary biomarkers identified in the included studies is presented in Table 5. This includes their biological role, clinical utility, limitations encountered, and suggested research directions.

Previous reviews have explored the diagnostic or prognostic role of salivary biomarkers; however, most combined diverse study designs, including observational, case-control, or in vitro studies, limiting the strength of their conclusions. In contrast, our systematic review is the first to synthesize only randomized controlled trials (RCTs), providing higher-level evidence on the therapeutic prognostic value of salivary biomarkers in both oncological and infectious diseases. Additionally, unlike prior reviews that were disease-specific (e.g., oral squamous cell carcinoma, COVID-19), our work provides a comparative perspective across a broader clinical spectrum. We further enhanced the methodological rigor by applying GRADE criteria to assess the certainty of evidence and by presenting a summary table that integrates biological significance, clinical utility, and future research pathways for each biomarker. These contributions offer a more robust, clinically oriented, and evidence-based foundation for future translational studies.

Despite the strengths of this systematic review—such as the use of a registered protocol, adherence to PRISMA guidelines, and exclusive inclusion of randomized controlled trials—several limitations must be acknowledged.

First, although the studies included were RCTs, six of them presented a high risk of bias (31, 33, 37, 42–44), particularly in aspects related to blinding, deviations from intended interventions, or incomplete outcome data. This contributes to uncertainty in the overall quality of evidence, as confirmed by the GRADE assessment, which rated the certainty as low due to concerns about risk of bias and inconsistency.

The high risk of bias across several studies, especially regarding inadequate blinding and deviations from intended interventions, undermines the internal validity and may lead to overestimation or underestimation of biomarker effects. In addition, the presence of serious inconsistency—reflected in the heterogeneity of effect sizes and variability in direction of outcomes—further reduces the confidence in the aggregated results. This inconsistency is likely due to differences in study populations (oncological vs. infectious), types of interventions (pharmacological, behavioral, photobiomodulation), biomarker types (e.g., IL-1β, IL-6, cortisol, nitrite), and sample collection protocols. Without subgroup analyzes or meta-analytic synthesis, it remains difficult to disentangle whether these differences represent true biological variation or methodological artifacts. Therefore, the combination of high risk of bias and inconsistency identified in the GRADE assessment justifies downgrading the certainty of evidence and underscores the need for more robust, standardized, and homogeneous future research.

Second, the sample sizes varied widely, ranging from 14 to 322 participants, and follow-up periods were often short or inconsistently reported. This heterogeneity reduces the ability to draw firm conclusions about the duration and clinical stability of salivary biomarker changes.

Third, differences in saliva collection and analysis methods—including use of stimulated vs. unstimulated saliva, variability in ELISA kits or multiplex systems, and timing of sampling—limit cross-study comparability and may introduce methodological bias (52).

Furthermore, in patients with head and neck cancer, high-dose radiotherapy is commonly used and may lead to significant salivary gland dysfunction, including xerostomia and compositional changes in salivary secretions. These alterations can compromise the stability, concentration, and detectability of bioactive molecules, such as cytokines and cortisol, potentially limiting the reliability of salivary biomarkers for prognostic purposes in this subgroup. Consequently, the applicability of saliva-based diagnoses in irradiated populations remains a clinical and methodological challenge that warrants specific investigation (53).

Fourth, although the review included both oncological and infectious diseases, the distribution of studies was even, with a predominance of cancer-related trials. Only two studies focused on HIV (34, 43) and one on tuberculosis (36), leaving viral, bacterial, and parasitic infections underrepresented. Therefore, the generalizability of results to the broader spectrum of infectious diseases is limited.

Finally, none of the included studies evaluated longitudinal prognostic trajectories beyond 12 months, and no study reported the cost-effectiveness or feasibility of implementation of salivary testing in routine clinical care, which limits the current clinical translatability of these findings (53).

The results of this review highlighted the potential of salivary biomarkers in the prognosis and monitoring of oncological and infectious diseases. However, the translational gap between research findings and clinical implementation remains wide, highlighting several priority areas for future research.

First, future studies should aim to validate salivary biomarkers in larger, multicenter RCTs, with adequate statistical power, standardized protocols for saliva collection and biomarker analysis, and longitudinal follow-up extending beyond 6–12 months. This will help determine whether changes in cytokine or cortisol levels are sustained over time and whether they correlate with clinical outcomes such as disease remission, relapse, or complications (54).

Second, comparative research is needed to evaluate the diagnostic and prognostic performance of salivary biomarkers vs. blood-based counterparts. If saliva proves to be equally or more sensitive, its application in non-invasive monitoring, particularly in low-resource settings or outpatient care, could be transformative (55).

Third, future investigations should include a broader spectrum of infectious diseases, such as hepatitis, COVID-19, or parasitic infections, where inflammatory and endocrine responses may be measurable in saliva. The inclusion of diverse populations, including pediatric and immunocompromised patients, will improve external validity and enhance the relevance of findings across healthcare contexts.

Fourth, attention should be given to the development of point-of-care (POC) salivary diagnostic devices, which could allow real-time monitoring of biomarker changes without requiring laboratory infrastructure. Advances in biosensors and microfluidics may soon enable rapid, portable saliva testing, revolutionizing the way clinicians track disease progression (56).

Finally, research on the cost-effectiveness, acceptability, and implementation barriers of salivary biomarker testing is urgently needed. Without this data, even highly accurate tests may fail to be adopted in clinical practice (57).