Investigating the potential association between tattoos and lymphoma: an exploratory systematic review and meta-analysis

Summary

Background

The tattooed population has risen significantly over the last few decades, leading to increased scrutiny into potential side effects. The recent publication of scientific articles linking tattooing to lymphoma has led to a systematic review being conducted to investigate if an association exists.

Methods

Following the PICO framework, we formed a systematic review comparing tattooed to non-tattooed adults belonging to any region of the world and investigated the likelihood of lymphomagenesis. The protocol was pre-registered on PROSPERO (ID: CRD42024586505). Relevant studies were searched for in PubMed, Web of Science, Embase, Google Scholar, and CENTRAL on 10/09/2024 and updated on 16/07/25. The inclusion criteria consisted of primary studies, including observational studies and case reports which investigated the association between tattoos and non-Hodgkin lymphoma. Exclusion criteria were publications involving subjects under 18 and non-English papers. Data extraction was performed using published numbers from individual papers after requesting raw data. Study quality was assessed using ROBINS-E, and evidence certainty using GRADE. Outcomes assessed were any odds/risk/incidence ratios that associated tattooing with non-Hodgkin lymphoma.

Findings

A total of four observational studies, totalling 17,941 participants (2485 cases and 15,456 controls) and three case reports were identified. None of the included studies demonstrated a statistically proven link between lymphomagenesis and tattooing. ROBINS-E showed relatively low bias for our three included studies; however, the certainty of our evidence is low due to the lack of high-quality studies with similar methodologies. The meta-analysis conducted for non-Hodgkin's lymphoma, with subtypes follicular lymphoma and diffuse large B-cell lymphoma, produced odds ratios of 1.01 (95% CI 0.82–1.24), 1.01 (95% CI 0.77–1.33) and 0.89 (95% CI 0.54–1.46), respectively.

Interpretation

No significant association was found between tattooing and lymphoma. Due to limitations in the data quality and lack of standardised measurable outcomes, further high-quality research is needed.

Funding

There was no funding for this study.

Research in context.

Evidence before this study

A preliminary search in pubmed and google scholar was performed with no start date limitations to September 2024, to help identify any studies which may prove relevant to our research question. Our search term involved various terms associated with the practice of tattooing (“Tatooing”, “Body Art”) separated with the term “OR” and various terms associated with lymphoma diagnosis (“lympoma”, “lymphamatoid”), again using the term “OR”. These two topics were then combined using the term “AND”. The preliminary search identified 3 case-control population based studies, including one published as an abstract. There were no systematic reviews or meta analyses identified as part of our preliminary search. The largest of the three studies suggested that tattooing was a risk factor for malignant lymphoma whilst the other articles demonstrated the opposite.

Added value of this study

We found four studies that met the inclusion criteria, collectively providing evidence on whether there is a risk of lymphoma from this procedure. We found that the current strength of evidence is low, and an increased risk or the possibility of no risk cannot be determined thus far.

Implications of all the available evidence

This systematic review demonstrates that there is inadequate existing evidence to suggest a link between increased risk of lymphoma and the practice of tattooing. Further population based studies would be beneficial to supplement the existing evidence, allowing a true estimate of effect to be estimated.

Introduction

Non-Hodgkin lymphoma (NHL) is the most common haematological malignancy in the world, constituting approximately 90% of all lymphoma cases.1, 2, 3 According to the GLOBOCAN database from the International Agency for Research on Cancer of the World Health Organization, there were 553,010 new cases of NHL worldwide in 2022; nearly 9000 more new cases than in 2020.^4,5 In spite of the significant disease burden, the aetiology of NHL remains poorly understood.¹ Several risk factors have been identified including environmental exposures, viral infections, genetic predisposition, and advanced age.^6,7 An environmental risk factor of particular interest is benzene exposure; benzene is a component of aromatic amines in tattoo ink and it has been demonstrated to have a dose-dependent relationship with lymphomagenesis.⁷ With the rise in popularity of tattooing, the increased exposure to benzene derivatives may pose a significant global health risk for future increased incidence of NHL.⁸

Tattoo inks have previously been described as cocktails of organic and inorganic colour pigments mixed with byproducts from the pigment synthesis and stabilizing additives.⁹ The composition of these inks recently came into the view of public policy with the European Chemicals Agency (ECHA) placing a restriction on thousands of hazardous chemicals found in tattoo inks and permanent cosmetic products.¹⁰ There is evidence to suggest that tattoo pigment can migrate to regional and distant lymph nodes in the body, organs which are highly sensitive to carcinogens.11, 12, 13 Ink spread to distant lymph nodes therefore indicates the possibility that various components in tattoo inks interact with the lymphoid system. As such, it is important to explore the possibility for epidemiological risk associated with this cosmetic procedure.

To date, there are no published articles which systematically quantify the association between tattoos and NHL. Despite the introduction of regulations by ECHA in recent years, there is still no international consensus or strong evidence regarding the risk for NHL development in the tattooed population.¹⁰ Given the high proportion of the general population who are tattooed, an accurate assessment of the NHL development risk associated with tattoos and the procedure's contribution of burden to NHL is imperative. This study is an exploratory meta-analysis quantifying the association between tattoo ink exposure and NHL. This study also included assessment of the quality of the available literature and an assessment of age-dependent risk.

Methods

We conducted our exploratory study following Cochrane's guidelines for systematic reviews and meta-analysis.¹⁴ The protocol was registered and published on the Prospective Register of Systematic Reviews (CRD42024586505) a priori.¹⁵ As part of a priori registration, we sought to conduct a systematic review alone, with meta-analysis only done if there was enough relevant data.

Eligibility criteria

To form the clinical question, the PICO framework (population, intervention, control, outcome) was utilised.¹⁶ All available experimental and observational studies which assessed lymphoma as an outcome from tattooing intervention with adult participants were included in this review, including case reports which identified lymphoma within the region of tattooing exposure. Opinion or review articles and non-human studies were part of our exclusion criteria, along with any studies in which participants were under 18. Any date and region of publication were included, and any non-English publications were excluded as authors could not accurately identify and appraise those articles.

Search strategy and information sources

Our search strategy was developed with the help of a medical librarian. The search strategy involved combinations of Medical Subject Headings (MeSH) terms and free text, with assistance from Boolean logical operators. References cited in the included articles were also screened for relevance. A preliminary search was performed beforehand during September 2024 within the Medline and Embase databases, identifying three published studies including one published abstract before the final systematic search. The conclusive search was conducted within the following databases: Medline, Embase, CENTRAL, Science Citation Index, and Google Scholar on the 16th July 2025. Grey literature and references from included studies were searched, and any appropriate articles were also included. The authors have provided an example of a Medline (OVID SP) search strategy (Appendix 1), which has also been adopted for other databases.

Data analysis

Search results were entered into Rayyan, an online reference collation and screening platform.¹⁷ Duplicates were removed manually based on title and abstract. Study screening was performed independently by two reviewers (TM, JX), and a third independent reviewer (JF) was consulted for any conflict over which studies were included. The screening process was reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram (Fig. 1).¹⁸ Authors of eligible studies, which were published as abstracts, were contacted to try and retrieve their full methodology and data set. The data from the studies was independently extracted by two authors (TM, JX) and exported into Microsoft Excel.¹⁹ The primary outcome for the extraction was a risk/odds ratio (using 95% confidence intervals) for NHL after tattooing. Secondary outcomes included risk of Hodgkin lymphoma (HL) and risk of lymphoma overall. Information for year of publication, type of study, lymphoma subtype analysis, population demographics were also recorded and analysed.

Fig. 1.

PRISMA flowchart showing the process of article identification within a systematic review of the association between tattoos and lymphoma. Search dated 16th of July 2025.

Study of risk of bias and certainty of evidence

To evaluate the risk of bias of studies, two authors (TM, JX) independently used Cochrane's ROBINS-E (Risk Of Bias In Non-randomised Studies—of Exposure) to produce an overall score.²⁰ A third independent reviewer (JF) was consulted for any conflict in scores. The Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) tool was used to evaluate the methodological quality of individual studies for NHL.²¹ A decision was made between the authors that the minimal clinically important difference (MCID) threshold for the odds/risk ratios should be 0.1, which will satisfy the imprecision section within GRADE. This threshold was chosen because any increased risk of lymphoma due to tattoo exposure is a public health concern. The inconsistency domain was decided based on I² values between the studies used for meta-analyses for NHL. An I² value < 40% was deemed as low, 30%–60% moderate and anything greater than 60% as considerable.

Synthesis methods

Data was later collated into a summary table (Table 1), with results and outcomes discussed. Forest plots were created using a random effects model in appropriate computer software (RevMan 5.4).²² Adjusted odds ratios were used where possible for any calculations to provide a more accurate estimate of the effect. If adjusted ratios were not feasible to use, odds ratios were calculated using raw numbers published in respective studies.⁹ Any NHL subtypes with data available for comparison between all the studies were used for the forest plot analysis. The Higgins I² statistic, formulated using RevMan 5.4 was used to assess variability between studies.^22,23

Table 1.

Summary of evidence and GRADE table assessing certainty of evidence between all studies for the purposes of evaluating the risk of tattoos and NHL.

Study characteristics							Certainty of evidence
Study	Type of study	Number of participants and location		Study aim	Effect measure		Risk of bias	Indirectness	Inconsistency	Imprecision	Publication bias	Overall GRADE
McCarty et al., 2024	Case Control	n = 9020 (Utah, United States)		To assess any association between tattoos and all haematological malignancies	Odds Ratio (OR)		Not serious	Seriousa	Low	Very Seriousb	Undetected	⊕◯◯◯ Very Low
					HL	NHL
		Cases n = 820	Controls n = 8200		0.66 (95% CI 0.36–1.21)	0.83 (95% CI 0.61–1.11)
Nielsen et al., 2024	Case Control	n = 5571 (Sweden)		To investigate the association between tattoo exposure and overall malignant lymphoma including subtypes	IRRc for Lymphoma	OR for NHL	Not serious	Seriousd	Low	Seriouse	Undetected	⊕⊕◯◯ Low
		Cases n = 1392	Controls n = 4179		1.21 (95% CI 0.99–1.48)	1.12 (95% CI 0.95–1.33)
Warner et al., 2020	Case Control	n = 1518 (British Columbia, Canada)		To assess whether tattoos increase the risk non-Hodgkin's lymphoma or Multiple Myeloma	OR for NHL		Not serious	Seriousf	Low	Very Seriousb	Undetected	⊕◯◯◯ Very Low
		Cases n = 737	Controls n = 781		1.05 (95% CI 0.66–1.62)
Clemmensen et al., 2025	Case Control	n = 66 (Denmark)		To study the potential association between tattooing and certain types of cancer including skin cancer, bladder cancer and lymphoma as a dual case control and cohort study	Case Control Hazard Ratio (HR) for Lymphoma				GRADE not applicableh
		Cases n = 32	Controls n = 32		1.35 (95% CI 0.69–2.71)
	Cohort	n = 2367			Not possible to measure Hazard Ratio for Lymphoma within Cohort Studyg
		Cases n = 6	Controls n = 2361

^aStudy outcome based on all haematological malignancies including both myeloid and lymphoid malignancies.

^bBased on a minimal clinically important difference (MCID) threshold of 0.10 OR either side of null effect with confidence intervals crossing both upper and lower thresholds.

^cIncidence rate ratio.

^dStudy outcome based on all lymphoma subtypes including Hodgkin's lymphoma.

^eBased on a minimal clinically important difference (MCID) threshold of 0.10 OR either side of null effect with confidence intervals crossing upper threshold using calculated unadjusted odds ratios.

^fStudy outcome based on both multiple myeloma and non-Hodgkin's lymphoma.

^gOutcome measure not measurable due to inadequate case numbers.

^hCertainty of evidence not applicable due lack of NHL measurement.

Role of funding source

There was no funding for this study.

Results

The literature search identified seven primary articles, within the chosen databases from 474 identified records (Fig. 1). Four of these sources were case–control studies, and the remaining three articles were case reports (Tables 1 and 2).

Table 2.

Identified case reports condensed into a table with patient, lymphoma and tattoo characteristics.

Case report	Patient characteristics			Tattoo characteristics		Lymphoma subtype
	Sex	Age at presentation	Medical history	Exposure time	Colour of tattoo
Kluger et al., 2024	Male	77	Hypertension Hyperuricemia Hiatus Hernia Basal Cell Carcinoma	Described as “old”	Black	Primary cutaneous marginal zone B cell lymphoma
Kluger et al., 2022	Female	28	No medical history	10 y	Multicoloured	Primary cutaneous follicle centre lymphoma
Sangueza et al., 1992	Male	54	No medical history	20 y or longer	Unknown but red pigment present	Monoclonal large B-cell lymphoma

The number of publications on this topic has risen dramatically over the last four years, with no primary studies arising before the year 2020. Case reports on this topic have also been very limited and have only been reported when superficial lymphoma originates in the region of the tattoo (Table 2).

The three case reports identified and examined in this review diagnosed varying types of NHL in the region of a tattoo.24, 25, 26 No single ink colour was reported more than others. It is therefore not possible to identify a trend with any specific colour. With the limited number of documented presentations, all the papers showed an exposure time of 10 years or longer, rather than immediately after tattoo placement. One of the case reports published by Arminger et al. diagnosed a 32 year male with histiocytic lymphoma at the site of the tattoo but was later reclassified as a pseudolymphomatous reaction by Blumental et al., so was excluded from our review after screening.^27,28

Three of the identified case control papers published by McCarty et al. (2024; Utah, USA), Nielsen et al. (2024; Sweden) and Warner et al. (2020, British Columbia, Canada) were used as part of our exploratory meta-analysis.^9,29,30 The study by Clemmensen et al. (2025, Denmark) was excluded for the meta-analysis due to differences in study design.³¹ The location of each individual study and population size of each study is demonstrated in Fig. 2.

Fig. 2.

World bubble map demonstrating study locations with size of bubble proportional to study size.³²

Certainty of evidence analysis was performed using the GRADE approach (Table 1) recommended by the Cochrane Handbook of Systematic Reviews.²¹ The reviewers used the ROBINS-E tool to determine the initial certainty of evidence (Fig. 3).²⁰

Fig. 3.

Risk of bias table demonstrating assessment of risk between all 3 studies using ROBINS-E (Risk Of Bias In Non-randomized Studies–of Exposures) created using the ROBVis tool.

The indirectness within GRADE analysis for the majority of the papers available, raised concerns due to the substantial differences in outcome measure. None of the papers directly addressed the specific question to quantify the association between tattoo ink exposure and non-Hodgkin's lymphoma. Imprecision was downgraded on all the papers based on our choice of minimal clinically important difference (MCID) threshold, which is a direct result of the confidence interval breadths. Inconsistency was deemed as low across all studies, as an I² value of 37% was achieved between all studies. Similarly, publication bias was not measured because a funnel plot was not appropriate. Overall, the certainty of evidence was of a lower quality, with indirectness and imprecision being the primary contributing factors Table 1.

The summary of the evidence for each specific study (Table 1) suggests that there is not enough data to show a correlation between tattoos and lymphoma at a glance. All studies fail to reject the null hypothesis that tattooing did not affect the incidence of NHL. Additionally, it is worth noting that comparing point estimates using adjusted data between studies proved difficult due to inconsistent effect measures and study aims. For example, Nielsen et al. (2024) provides an incidence rate ratio, which is the ratio of odds to incidence per person time from an incidence density sampling method, whilst the other two comparable case–control studies involving NHL used an odds ratio based on more traditional sampling methods⁹ Warner et al. (2020) chose to study associations with non-Hodgkin lymphoma (NHL) but not Hodgkin lymphoma (HL), a separate outcome from the other two case control studies, which explore both Hodgkin and non-Hodgkin lymphomas.³⁰ Therefore, odds ratios were calculated in an unadjusted form for Nielsen et al. (2024) to provide statistical uniformity using the raw numbers attached in the Supplementary Files of combined effects for purposes of any meta-analysis.⁹ Lastly, the study by Clemmensen et al. (2025) provided a hazard ratio and was a twin case control study so the data will not be comparable in the meta-analysis but provides useful information regarding the pathogenesis of lymphoma from tattoos.³¹

When data for non-Hodgkin's lymphoma was isolated from each study for forest plot Fig. 4a using an inverse variance random effects model, the pooled data demonstrated that there was insufficient evidence to demonstrate any association between tattoos and NHL. The forest plot showed a pooled odds of 1.01 (95% CI 0.82–1.24) and low heterogeneity between papers with an I² statistic of 37%. This result was similar for the two subtypes explored: Follicular lymphoma (FL) and diffuse large B cell Lymphoma (DLBCL) Fig. 4b and c. These subtypes had a pooled OR of 1.01 (95% CI 0.77–1.33) and 0.89 (95% CI 0.54–1.46), respectively.

Fig. 4.

Forest plots demonstrating the combined estimated effect of tattoos on NHL (a, top), FL (b, middle), DLBCL (c, bottom) calculated by the inverse variance random effects model using confidence intervals from the primary studies created in Revman 5.4. Totals were not included as adjusted odds were used where possible to provide a more accurate point estimate.

Equivalence testing was performed for pooled NHL values. The authors chose predefined delta values of 0.1 as a maximum statistically equivalent effect. Pooled NHL was chosen for equivalence testing only as the sample sizes for the subtypes were much smaller and therefore inappropriate for equivalence testing. The equivalence testing showed both non statistically different and non-statistically equivalent results for NHL, since the upper boundaries cross the line of equivalence (Fig. 5).

Fig. 5.

Equivalence testing for pooled NHL with OR of 1.01 (90% CI 0.85–1.20) using a lower and upper equivalence boundary of Δ+0.1 and Δ−0.1.

Sensitivity analysis

Sensitivity analysis was performed using data from the Clemmensen et al. study under the assumption that all the patients who developed lymphoma were NHL cases. Raw unadjusted OR figures were trialed, since adjusted OR data was not available for standardised comparison. We also utilised the adjusted HR data under the premise that it is closer to the true adjusted OR as an adjacent sensitivity analysis. This showed that the pooled OR was 1.03 (95% CI 0.89–1.20) and 1.04 (95% CI 0.87–1.23) for use of unadjusted OR and HR respectively. The added study included an extra 10 tattooed cases to the pooled analysis and therefore provided the least weighting out of all the utilised studies. Under the new assumptions, the results fail to reject the null hypothesis.

Demographic analysis

The percentage of male and female lymphoma cases remained relatively the same across all three studies apart from the Danish study which showed a greater skew towards the male demographic (Table 3). Nielsen had a markedly lower percentage of individuals over the age of 60 which formed part of their study likely due to the exclusion of patients older than 60 years of age.⁹ However, controls could still become patients during the study if they developed lymphoma which accounts for the number of patients in the over-60 age bracket. The age ranges between studies therefore remained inconsistent for the meta-analysis. The prevalence of tattooing in NHL groups in McCarty et al. was 11%, and the overall prevalence of tattooing seemed to differ vastly amongst regional populations.²⁹

Table 3.

Patient age demographics for each study investigating the association between tattooing and lymphoma are shown as percentages.

Study	Demographics
	Tattooed (%)		Male (%)		Age (%)
	Case	Control	Case	Control	20–49		50–59		60+
					Case	Control	Case	Control	case	Control
McCarty et al., 2024	12	15	56	53	26	26	18	18	56	56
Nielsen et al., 2024	21	18	55	53	45	46	45	48	11	12
Warner et al., 2020 (NHL only)	6	6	59	54	20	27	25	21	56	52
Clemmensen et al., 2025 (Case Control)	31	32	66	44	Not Recorded

Out of all the patients who had lymphoma, the prevalence of NHL in the two overall lymphoma studies was 86% for McCarty and 76% for Nielsen et al. (2024).^9,29 When McCarty et al. (2024) stratified their data between 20 and 60 years of age, the odds ratios for NHL overall was 1.03 (95% CI 0.71–1.49), lower than the reported 0.83 (95% CI 0.61–1.11) in the 19–79 year old group.²⁹ It was not possible to include Warner et al. (2020) 22 in the 20–60 analysis as this was not reported, whilst Nielsen did not publish an adjusted NHL point estimate for the same demographic.^9,30

Risk of bias

The analysis of ROBINS-E on these studies resulted in relatively low concerns for bias overall Supplementary Data.²⁰ All authors attempted to reduce the risk of confounding factors by adjusting for confounders where available, though these were not consistent between studies. Common confounders included age, sex, and education.

Concerns were raised regarding the lack of control matching for the study provided by Warner et al. (2020).³⁰ However factors such as age, sex, family history and other potential confounding factors were adjusted for later. Clemmensen did not adjust for potential confounding factors such as smoking, occupation and socio-economic status as part of their case–control portion of their study, despite negating genetic factors through the use of a twin based population data set.³¹ The full data set was also requested from the authors of the three studies used for meta-analysis. There was no response from Warner after two attempts to contact them, and the other two authors suggested that this was not possible due to confidentiality.

Discussion

To our knowledge, this is the first systematic review and meta-analysis to investigate the association between tattoo exposure and the risk of non-Hodgkin lymphoma (NHL). This exploratory review was conducted due to the rise in news reports suggesting an increased risk. Prior to 2020, no primary studies had addressed this topic, and the available literature was limited to a small number of case reports suggesting a potential link. The recent increase in research may be attributed to several factors, including a general rise in scientific output, the growing prevalence of tattoos, and recent European Union regulations concerning the chemical composition of tattoo inks.¹⁰ This increase has been demonstrated by the recent publication of four large observational studies within the span of 4 years, though none of these studies were unable to individually demonstrate association.

In the exploratory meta-analysis of these three identified studies, 15,574 individuals were included, with 3260 diagnoses of NHL. The pooled analysis revealed not enough evidence to demonstrate increased risk of overall NHL associated with tattoo exposure, with consistent findings across all included studies, regardless of design and the same within our subgroup analysis. Equivalence testing for NHL with a meaningful equivalence boundary of 0.1 showed that an effect cannot be determined as shown by both non-equivalent and non-significant results. Since a raw unadjusted odds ratio was used from the study by Nielsen et al. there is a risk that the forest plot does not provide a true pooled effect.⁹ This is compounded by the fact that the Swedish study is the heaviest weighted in the meta-analysis, potentially skewing results. Our sensitivity analysis utilising the study by Clemmensen et al. resulted in a pooled odds ratio of 1.03 (95% CI 0.89–1.20) and 1.04 (95% CI 0.87–1.23) using unadjusted OR and HR respectively, failing to reject the null hypothesis in either circumstance. Other meta-analysis methods considered include the Mantel-Haenszel method, which was not performed due to the involvement of large-scale population studies using adjusted data. The Peto methodology was also considered, however with the natural heterogeneity of tattooing especially in various locations and populations, this was excluded to prefer a model which allowed use of random effects.

The certainty of evidence studies (COE) were mostly poor, as determined using the GRADE methodology.²¹ In most situations, the studies deemed as having a very low COE would be excluded; however, for our review, this would have provided inadequate data for meta-analysis. There is the hypothetical possibility that the presence of studies with a higher COE and also statistically significant results could have changed our overall conclusion by nature of excluding the studies with lower COE. In addition, our pooled point estimate uses unadjusted odds ratios for the study by Neilson (2024) despite our preference for adjusted ratios.⁹ This was done as Nielsen et al. used IRR as their effect measure, which was not ideal for a standardised comparison.⁹ The only study not included in the meta-analysis but within the review was Clemmensen et al. (2025) due to significant differences in study design. However, the twin study does provide a good understanding of the prolonged risk of tattoo exposure over time, although limited by small sample size.

Between all the studies there were only two statistically significant results based on tattoo size and exposure time. Clemmensen reported that a tattoo size larger than a palm resulted in an overall lymphoma HR of 2.37 (95% CI 1.11–5.06), contradicting the reported IRR of 1.14 (95% CI 0.86–1.53) in the study by Nielsen et al.^9,31 It is worth noting the reported HR has not been adjusted for environmental exposures such as occupation which may indicate some confounding compared to the Swedish study. There may also be variability between measurement of exposure based on palm sizes between the authors and tattoo composition in both populations leading to differing results. Exposure time analysis by Nielsen et al. (2024) suggested that the incidence rate ratio was highest within the first two years 1.83 (95% CI 1.05–3.21) and after 11 years 1.23 (95% CI 0.98–1.53) for overall lymphoma.⁹ A possible explanation for the increased short-term risk described by Nielsen et al. (2024) could potentially arise from an increased initial inflammatory response.^9,33 However, further research is necessary to determine whether this increased risk is significant. In contrast, data from Warner et al. indicated that an exposure duration of less than 25 years resulted in an odds ratio for NHL of 0.92 (95% CI 0.48–1.71), and McCarty reported a 20 year NHL risk of 0.60 (95% CI 0.34–1.05).^29,30 Nevertheless, these timeframes are not directly comparable, as the inflammatory response likely contributes to the increased risk shortly after the procedure and may diminish over time. It is also worth mentioning that the two studies that did find some significance within their results are both from the northern European region, suggesting that there could be unknown confounding factors present which have yet to be identified.

Interestingly, studies that did not establish an upper age limit for participants over 60 years reported that more than 50% of their patients fell into this age group.^29,30 This trend, as seen in Warner et al., is primarily attributed to the prevalence of NHL cases and the cumulative impact of environmental damage over a lifetime.³⁰ As a result, this age range may not be ideal for accurately assessing the true risk of NHL associated with tattoo exposure, as potential confounding factors could accumulate, reducing the reliability of the data.

From the case studies identified, all cases reported were instances of cutaneous presentations of NHL. There was no overall particular trend between reports. It is also hard to infer any data from case studies, as cases were picked up only if the lymphoma lesion was found on the site of the tattoo. This ignores the possibility that lymphoma may present elsewhere in the body secondary to tattoo application. As previously evidenced in rodent studies, tattoo ink can distribute to local and regional lymph nodes.¹² Therefore, it is possible that patients can have distant or regional presentations of lymphoma but lack direct association to the tattoo due to lymphomagenesis away from the site of the tattoo. Nonetheless, current data does not yet confirm these hypotheses within human populations.

This is the first review of studies evaluating the risk of lymphoma from tattoo exposure and shows an increasing number of publications within recent years. The evidence base on this topic is currently in its infancy and warrants further population-based studies to contribute to a more precise lymphoma risk. Although only three studies were eligible for meta-analysis, the pooled sample included 15,574 individuals, spanning three distinct populations: British Columbia (Canada), Utah (USA), and Sweden, and one included twin study for the systematic review from Denmark. These geographic locations introduce potential variability in effect estimates due to regional differences in tattoo ink composition and levels of sunlight exposure factors that may interact with certain ink constituents. However, current research on these interactions remains limited, making it difficult to draw robust conclusions about the carcinogenic risk of tattooing. Secondly, two of the included studies predominantly sampled older individuals, which may limit the external validity and generalisability of our findings to generalised populations. As the population age increases it may be difficult to define whether the lymphoma is due solely to tattoo exposure or other environmental damage. The included study by Neilson et al. (2024), which had the strongest effect, limited their study to under 60s due to variability in effect in patients older than this.⁹ However, we were not able to compare and analyse this as there was unavailable moderator data for age at the individual level across two of the studies for NHL.^9,29

There is also potentially variability in classification of histological subtypes of NHL across the studies, Nielsen et al. (2024) was the only study providing classification subtypes in their study. Without transparency for which subtypes were classified as for the other studies, it could affect accuracy of both pooled analysis and direct comparisons between publications.

A further limitation to our meta-analysis comes from the use of non-adjusted data for one out of the four studies, therefore limiting the quality of pooled estimates. The study, published by Nielsen et al. additionally provides the largest weight for the meta-analyses. This potentially skews the true combined effect as both environmental and genetic factors could not be accounted for in the highest-powered publication currently available. The results from our meta-analysis are part of an exploratory review into a rapidly developing field rather than as definitive evidence. Further studies with heterogenous effect measures in the form of adjusted OR or IRRs would be beneficial in providing access to a more accurate pooled estimate.

Our study potentially has clinical implications. Tattoo prevalence has been on the rise in recent years, with increased concern only recently.⁸ The results of this systematic review suggest that there is not enough evidence to link tattoos to NHL. Furthermore, due to the exploratory nature of the meta-analysis and the lack of heterogeneous studies, it is difficult to draw robust conclusions on whether tattoos could increase the risk of lymphomagenesis. We would encourage authors to publish further in vivo studies and population-based studies in various other regions of the world to help broaden the evidence base available for this topic.

Contributors

TM and JX carried out all stages of the review including conceptualisation, data curation, formal analysis, investigation, methodology, project administration, resources, software, supervision, validation, visualisation, writing—original draft, and writing, review & editing. Jack F and IZ assisted in data curation, formal analysis, investigation, methodology, and revision of the manuscript. John F assisted in validation and supervision of the project, data verification, and revision of the manuscript.

TM, JX to be considered joint first authors.

Data sharing statement

All data generated or analysed during this study will be available.

Declaration of interests

All authors declare no competing interests.