Idiopathic granulomatous mastitis: A scoping review to determine the quality and geography of the evidence base Authors: Daniel Ahari, Leah Argus, Rute Castelhano, Shaneel Shah, Goonj Johri, Emma MacInnes, Christopher Darlow, Iain Lyburn, Nisha Sharma, Vijay Sharma, Rudresh Shukla, Roisin Bradley, Nazina Arafin, Cliona C Kirwan Published: 2026 ABSTRACT AbstractIdiopathic granulomatous mastitis is a rare chronic inflammatory breast condition with uncertain aetiology, inconsistent diagnostic pathways, and variable treatment practices. This scoping review aims to map the breadth, geographical origins, and methodological quality of the current published literature, thereby evaluating its strength and applicability in guiding clinical practice in the United Kingdom and similar healthcare systems.A literature search on March 9th 2023 revealed 3208 records: 225 primary studies and five systematic reviews met inclusion criteria. The primary studies reported on 13,062 patients from 39 countries, with Turkey represented most. Studies were predominantly retrospective and single-centre, with small sample sizes. Fifteen authors appeared on multiple papers; assessment of these publications showed that 24 papers, encompassing 2242 patients (17.2% of all patients), may have overlapping cohorts.Of the 225 primary studies, 175 (77.8%) reported at least one treatment or intervention. When appraised using the Oxford Centre for Evidence Based Medicine criteria for treatment benefit, most evidence (122 studies, 69.7%) was level four (e.g. case series). A further 50 studies (28.6%) provided level three evidence (e.g. non-randomised follow-up studies), and only three studies (1.7%) were randomised controlled trials, representing level two evidence. No studies could be categorised as level one (highest strength of evidence). The systematic reviews included 6262 patients across overlapping primary studies, and incomplete referencing in one review prevented accurate cross-matching between the reviews and with the primary studies.Overall, current evidence is geographically concentrated, methodologically weak, and insufficient to guide practice. High-quality prospective and randomised studies are urgently required.Keywords: Idiopathic granulomatous mastitis, Mastitis, Benign breast, Review Introduction Idiopathic Granulomatous Mastitis (IGM) was first described by Kessler and Wolloch in 1972 as a rare inflammatory breast condition presenting as a mass and characterised by histological granulomas that can be differentiated from other recognised forms of mastitis or fat necrosis [1]. Its exact pathogenesis is unknown, but many factors having been proposed as potential risk factors. These include infections, childbirth, oral contraceptive use, autoimmune disease, and diabetes mellitus [2]. The clinical presentation mimics acute bacterial infection, breast abscess, and even malignancy, hence there can be diagnostic delay. Given the lack of pathognomonic features, it is often a diagnosis of exclusion. In addition, treatment is highly varied, largely non-evidence based, and has low efficacy. Although there are numerous publications on the diagnosis and management of IGM, the quality of evidence is highly variable and potentially geographically biased. Any weighting of the current evidence base to specific geographical areas and thus ethnicities may reflect a true higher incidence in these countries, under-reporting or underdiagnosis in other regions, or even overdiagnosis in other healthcare systems where access to imaging and pathology are less consistent. In recognition of the lack of evidence relevant to western populations and healthcare systems, the Establishing National Guidance for Idiopathic Granulomatous Mastitis (ENIGMA) study was undertaken to determine the existing evidence base, how it applies to UK population, and clarify which research gaps exist - with the overarching aim of developing national guidelines for the diagnosis and management of IGM in the UK and similar populations and healthcare systems [3]. As part of this programme of work an in-depth scoping review was performed with the aim of determining evidence strength and highlighting evidence by gaps, through quantifying the range, quality, and geographical origins of the existing literature, with particular consideration of duplicate cohort publication. Methods A literature search was performed on March 9th 2023 for papers published between the date of the search and the preceding 20 years of records for Ovid MEDLINE®, Embase, and the Cochrane Central Register. The search was supported by the Royal College of Surgeons of England Library and Archives Team. The search terms included ‘granulomatous mastitis’, ‘lobular mastitis’, and ‘cystic neutrophilic granulomatous mastitis’ (CNGM), while excluding references to lactation or pregnancy (Supplement 1). Studies of human participants with IGM from all age groups and all stages of care, including diagnosis, management, and follow-up were included. Full-text primary research articles including case reports (n > 1), case series, cohort studies, trials, systematic reviews and meta-analyses were accepted. If full-text articles were not available through open access, The University of Manchester Library, Edge Hill University Library and NHS England OpenAthens, the study authors and publishers were contacted. Records were managed using the Rayyan software. Further details are reported in the ENIGMA protocol paper [3]. Our exclusion criteria were: studies not available in English, animal studies, cadaveric studies, conference abstracts or studies where the full-text was not available, non-original research (e.g. letters to the editor, literature reviews), trial or study protocols, studies where the intervention was not licensed by the Medicines and Healthcare Products Regulatory Authority (such as traditional Chinese Medicine), and studies investigating non-idiopathic granulomatous mastitis except for CNGM (i.e. tuberculosis/TB, sarcoidosis, IgG4-related disease). The results were screened by their titles and abstract by two independent reviewers (LA and SS), with a third reviewer (CK) resolving discrepancies. Thereafter, the papers’ full texts were reviewed by five independent reviewers (DA, LA, RC, SS, GJ), to determine whether the paper met the inclusion criteria. Additional backward citation search of any systematic reviews identified was also performed, wherein their referenced papers were cross checked against the screened results. Any previously unidentified eligible papers were then also included in the results. A data collection proforma was used to ensure uniform extraction and organisation of study characteristics and data. The proforma was designed using the Data Extraction for Complex Meta-Analysis (DECiMAL) guide and Template for Intervention Description and Replication (TIDieR) [4,5]. Data collection was then performed from the full texts of included papers. To assess relevance of the published literature in guiding clinical practice in the UK, data points collected included (but were not limited to) those pertaining to recency, origin, study design, sample size, and patient demographics. The level of evidence was classified according to the Oxford Centre for Evidence Based Medicine (OCEBM) [6]. Once studies were identified, the first and last authors of all included papers were also cross-checked for duplication. Papers sharing a first or last author were then examined in detail to assess potential overlap based on the data-collection centre(s), study period, and inclusion criteria. The scoping review was registered with PROSPERO (International Prospective Register of Systematic Reviews, CRD42023397427). Results Of 3208 records identified, after removal of duplicates, title and abstract screening, and full-text review, 225 papers met inclusion criteria. In addition, backward citation searching of five eligible systematic reviews identified five further papers, bringing the total to 230 (Fig. 1). PRISMA flow diagram demonstrating number of records identified, included, and excluded. Overall, 225 were primary studies and five were secondary syntheses (four of which had meta-analyses). Year and origin of publication There was a steady increase in the frequency of publications over the study period (Fig. 2). Of these studies, the earliest date of diagnosis for an included patient was 1970, with the most recent including patients diagnosed in 2022. The longest study spanned 33 years of data collection; the largest study (1309 patients) reported data over 31 months (October 2016 to May 2019). Of all primary studies included, 106 studies out of 225 (47.11%) did not report the relevant dates of data collection. Primary publications on Idiopathic Granulomatous Mastitis by year. Note, as the scoping review was conducted on March 9th 2023, the reported frequency for 2023 does not reflect the full year. The 225 primary publications originated from 39 countries (Fig. 3) with Turkey contributing the highest number of publications (n = 86, 38.2%), followed by the USA (n = 30, 13.3%), China (n = 23, 10.2), and Iran (n = 11, 4.9%). Almost all studies regardless of having multiple sites were performed in a single country (n = 221, 98.2%), with only 3 studies spanning two countries and one study which did not clearly state the country of origin. Primary publications per country. The vast majority of primary studies (196, 87.1%) were single site, with fourteen studies being across 2 sites; six across 3 sites; one across 4 sites, four across 5 sites. There was one study carried out across 9 sites, and one carried out over 22 sites. The five systematic reviews were published from 2017 to 2021; three were published in China, with one each in Iran and Spain. Study size In total, there were 13,062 patients across the 225 primary papers included in the scoping review (Fig. 4). Of these, only six studies included a cohort of over 200 patients, with 153 (68.0%) having cohorts of 50 or under, and 87 (38.7%) having cohorts of 20 or less. The median sample size was 30 (IQR 13 - 63). Sample size of each included primary study (n = 225). Specific sample sizes have been provided for the five largest studies. Risk of duplication Fifteen unique authors were identified as first and/or last author on more than one primary paper: twelve authors each on two papers, one author on three papers, and two authors on four papers (35 papers in total, encompassing 2643 patients). Of these fifteen authors, five (accounting for eleven papers) produced studies that were determined to have mutually exclusive cohorts, based on non-overlapping data-collection periods, different sites, or inclusion criteria. Therefore the 401 patients included in those eleven papers were found likely to be unique and not included in other papers. Thus, the remaining 24 papers included 2242 patients (17.2% of all patients), which may have been included in more than one paper (Fig. 5). Flowchart for assessment of patient cohort overlap. Systematic reviews and meta-analyses Five systematic reviews were identified in our literature search, four of which conducted a meta-analysis (Table 1). Backwards citation searching of the reviews, ensured that, when references were provided, all eligible papers were included in our review; five papers were identified this way (Fig. 1). A total of 128 papers were included over the five reviews, with a total sample size of 6292. This equivalent to 48.17% of the 13,062 patients included in the 225 primary papers. Table 1Description of systematic reviews.Table 1Study (Year)Meta-Analysis Performed?ObjectiveNumber of Studies (and Participants)Libraries SearchedTime PeriodIncluded StudiesMartinez-Ramos et al. [7] (2019)NoTo describe the most frequent clinical manifestations and treatments reported in the scientific literature.70 (3060)PubMed, EMBASE, Cochrane Library and Google ScholarMay 22, 2013 - 201870 studies, no description beyond this and several studies missing.Godazandeh et al. [8] (2021)YesTo assess the effectiveness of steroids on IGM, comparing surgery alone, surgery+steroids and steroids alone12 (559)PubMed, EMBASE, Cochrane Library and Web of ScienceStart of records - May 201912 studies: 5 case series, 6 cohort studies and 1 case control study.Ma et al. [9] (2020)YesTo assess the effectiveness of steroids and/or surgery; methotrexate; antibiotics; and observation on IGM.21 (970)PubMed, EMBASE, and the Cochrane LibraryStart of records – September 10 201821 case control studies.Lei et al. [10] (2017)YesTo assess the effectiveness of steroids and/or surgery on IGM with respect to complete remission and recurrence.15 (602)PubMed, EMBASE, Cochrane Central Register of Controlled Trials, and Medline2010 - 201515 studies - all non-randomised and mostly retrospective.Zhou et al. [11] (2019)YesTo evaluate the clinical effectiveness of the conservative versus surgical treatment for IGM.10 (1101)PubMed, EMbase, ScienceDirect, and Web of ScienceJanuary 1981 and April 201810 non-randomised, not otherwise specified studies included (n = 1101). Description of systematic reviews. However, we cannot say how many of these 6292 patients are represented in the 13,062 patients from the primary papers. Firstly, the systematic reviews each included overlapping sets of primary studies, i.e., the 128 studies in the reviews do not represent 128 unique publications. Secondly, one systematic review (Martinez-Ramos et al.) did not provide references for its 70 included papers [7], meaning that precise cross-matching and accurately assessing the overlap of primary papers and patients in the reviews with our dataset could not be accomplished. For these reasons, we report the 13,062 patients from primary studies as our main population and describe the systematic review populations narratively only, ensuring no double counting of patients between sources. Martinez et al. compared use of steroids, surgery and antibiotics in developed and 'non-developed' countries [7]. Seventy studies were included (n=3060), although only 20 references were provided and so multiple studies could not be identified for further analysis here. Fifty-six studies contained 50 or fewer participants. Turkey was the origin for 21 papers, 12 were from the USA, seven from Iran, four each from Egypt and India, three each from Japan, Spain, and South Korea, two each from Australia and China, and one each from Canda, Hong Kong, Israel, Mexico, Morocco, Oman, and the UK. Two studies also originated from multiple sites across Egypt and Morocco. Godazandeh et al. used meta-analysis with the aim of determining the efficacy of steroids, versus surgery, versus steroids and surgery [8]. They included twelve studies with between 11 and 77 participants, totalling 559 patients. All included studies were Middle Eastern or Asian in origin (Turkey being the most represented (n = 8), with one each for China, Iran, Jordan, and South Korea). Ma et al. used meta-analysis to assess the efficacy and safety of management options for IGM [9]. This included 21 case-control studies (n = 970), which contained between 8 and 206 patients, with only six studies including over 50 patients. Eight studies originated from Turkey, three from the USA, two each from Japan, South Korea and the UK, with one each from Brunei, China, France, and Israel. Lei et al. aimed to evaluate remission and recurrence rates for different IGM treatments via meta-analysis [10]. Fifteen non-randomised studies (n = 768) included use of medications, surgery, a combination, or observation. Sample sizes ranged from 2 to 53 patients. Seven studies originated from Turkey, with two from Iran and the USA, and one each from Israel, Japan and South Korea. Zhou et al. used meta-analysis to compare conservative and surgical treatment of IGM [11]. Ten studies with participant numbers ranging from 11 to 169 (n = 1101) were included; all were retrospective, and non-randomised. Five studies originated from Turkey, two from Iran, and one each from Japan, Saudi Arabia, and the USA. Levels of evidence The OCEBM defines what level of evidence a study may be as regards a specific clinical question. Systematic reviews and meta-analyses are usually considered the highest level of evidence (level one). To determine whether a diagnostic or monitoring test is accurate, the OCEBM recommends “systematic review of cross-sectional studies with consistently applied reference standard and blinding,” as level one evidence. To establish whether an intervention helps, the OCEBM recommends “systematic review of randomized trials or n-of-1 trials,” again as level one evidence [6]. Of all 225 primary studies, the largest group of study type was case series (n = 141), followed by cohort studies (n = 68). Most of these studies were performed retrospectively (197 studies, 87.6%). Of the retrospective studies, the majority were case series (n = 128, 56.89%) or cohort studies (68, 30.2%). Only 23 studies (10.2%) were performed as prospective studies, of which four were clinical trials (including three RCTs [n = 30, 36, 124] [[12], [13], [14]] and one non-randomised control trial [n = 78] [15]). The remaining studies used both prospective and retrospective data (four studies, 1.8%) or did not report directionality (one study). Of the 225 primary studies, 175 (77.8%) provided information on at least one form of treatment/intervention. Hence, when we apply the OCEBM ratings to these 175 studies for the question of ‘treatment benefits,’ most of the evidence is level four (single arm trials [n = 1], case-control studies [n = 8] and case-series [n = 113]; n = 122 or 69.7%). Cohort studies (n = 49, 28.0%) and non-randomised clinical trials (n = 1) are level three evidence (n = 50, 28.6%). Finally, the RCTs (n = 3, 1.7%) are level two evidence (Fig. 6). Bar chart demonstrating the OCEBM level of evidence for all 225 primary studies. The systematic reviews and meta-analyses included here fall short of the OCEBM standards, since none have included RCTs in their analysis. Given the inapplicability of OCEBM to the systematic reviews, a further analysis was carried out using AMSTAR 2, a widely used critical appraisal tool for systematic reviews and meta-analyses on healthcare interventions that contain randomised or non-randomised studies [16]. Multiple failures in the domains of AMSTAR 2 can demote overall confidence in a review's results from high to moderate, to low, and finally critically low. Sixteen domains are considered with this tool. Applying this to the four metanalyses, several weaknesses were identified, particularly around bias (Table 2), resulting in downgrading of the studies to moderate-to-low confidence. Table 2AMSTAR 2 appraisal of the four included meta-analyses.Table 2ItemAMSTAR 2 DomainGodazandeh et al.: Assessment and ReasoningZhou et al.: Assessment and ReasoningLei et al.: Assessment and ReasoningMa et al.: Assessment and Reasoning1Satisfaction of PICO (population, intervention, comparator, outcomes)YesPICO clearly definedYesPICO clearly definedYesPICO clearly definedYesPICO clearly defined2Protocol registration before commencementNoNot stated in reviewNoNot stated in reviewNoNot stated in reviewNoNot stated in review3Justification of included study typesYesHuman trials using steroids in confirmed IGM patients with recurrence ratesYesAll included were observational, specifically as no RCTs existedNoNo justification provided for not including RCTsYesAll included were case-control as no RCTs available4Comprehensive literature searchPartial yes>2 databases used and search strategy given; included studies' references not searchedPartial yes>2 databases used and search strategy given; included studies' references not searchedPartial yes>2 databases used and search strategy given; included studies' references not searchedPartial yes>2 databases used and search strategy given; included studies' references not searched5Eligibility assessment by ≥ two authorsYesClearly statedYesClearly statedYesClearly statedYesClearly stated6Data extraction by two authorsYesClearly statedNoNot clearly statedYesClearly statedYesClearly stated7List of excluded studies with individual justificationNoExcluded studies not listed; reasons for exclusion not givenNoExcluded studies not listed; reasons for exclusion not givenNoExcluded studies not listed; reasons for exclusion not givenNoExcluded studies not listed; reasons for exclusion not given8Adequate description of included studiesPartial yesInformation given on studies; missing information on follow-up timePartial yesInformation given on studies; missing information on dosagesPartial yesInformation given on studies; missing on dosagesPartial yesInformation given on studies; missing on dosages9Risk of bias assessment for included studiesPartial yesNewcastle-Ottawa Scale (does not assess selective reporting)Partial yesNewcastle-Ottawa Scale (does not assess selective reporting)NoRisk of bias not assessedPartial yesJadad score mainly used (does not assess selective reporting); GRADE score also used but results not discussed.10Reporting of funding for included studiesNoNot reportedNoNot reportedNoNot reportedNoNot reported11Appropriate meta-analytical methodsNoNo mention of using adjusted estimates or accounting for confounding; no justification for pooling dataNoNo mention of using adjusted estimates or accounting for confounding; no justification for pooling dataNoNo mention of using adjusted estimates or accounting for confounding; no justification for pooling dataNoNo mention of using adjusted estimates or accounting for confounding; no justification for pooling data12Consideration of effect of risk of bias on meta-analysisNoNot discussedNoNot discussedNoNot discussedNoNot discussed13Consideration of risk of bias when discussing resultsNoRisk of bias mentioned but impact not discussedPartial yesCautious interpretation of results; no explicit mention of impact of study design or NOS scoreNoNot discussedNoNot discussed14Explanation and discussion of result heterogeneityNoNot discussedNoSubgroup analysis performed but did not continue to explore sources of heterogeneityNoHeterogeneity reported but impact not discussedNoHeterogeneity reported but impact not discussed15Assessment and discussion of publication biasPartialEgger's test used; publication bias reported but not discussedYesBegg's test performed; no significant bias foundNoNot assessedYesEgger's test used; publication bias reported: no significant bias16Sources of conflict of interest or funding declaredYesNone declaredYesNo conflicts declared; funding (national and municipal) declaredNoNo conflicts declared; funding not mentionedYesNone declared AMSTAR 2 appraisal of the four included meta-analyses. Discussion Our scoping review demonstrated that while the number of publications on IGM is increasing over time, the data remains concentrated around the Middle East (particularly Turkey), Northern America, and China. Only 25 (0.19%) of the 13,062 IGM patients in the analysed literature were UK based, and 76 (0.39%) from Europe (excluding the UK). This may reflect a low prevalence but may also reflect underdiagnosis, with lack of awareness as a potential factor. The geographical bias of the existing literature means the current evidence base is unlikely to be generalisable to healthcare systems beyond those strongly represented in the literature, such as the UK and western Europe. However, beyond the geographical bias, the existing literature on IGM is mainly composed of low-quality evidence. Of the 225 primary papers we identified, most are single-centre, retrospective case series, with 153 studies including 50 participants or fewer and an additional 47 studies including 51 – 100 participants. Only 10.2% of studies were prospective in design, and case series were by far the most common study type (n = 141). The limited methodological quality of these studies - often lacking standardised diagnostic criteria, with highly variable management strategies and incomplete follow-up, means that they more closely reflect reports on local practice than generalisable evidence. The five systematic reviews commented on the benefits of potential treatments for IGM, however, given the lack of RCTs, none qualified as ‘level one’ evidence as per the OCEBM definition. Secondary analysis with AMSTAR 2 also revealed moderate to low confidence in the results. Nevertheless, we appreciate that the ability of a systematic review to answer a question hinges upon the strength of existing studies, and we therefore reaffirm some of the conclusions of the systematic reviews: that the existing literature lacks consensus on myriad issues such as definitions, treatment options, and treatment efficacy. This highlights the urgent need for unbiased high quality prospective studies and randomised control trials with clear inclusion and exclusion criteria, a priori power calculation, with clearly defined and quantifiable endpoints. Although our inclusion criteria specified that cohorts must include patients diagnosed with IGM, many publications did not define their diagnostic criteria and therefore may not have been truly idiopathic. For example, some of the countries where these publications arise are countries where TB is endemic, but less than half of the studies conducted microscopy and culture, and only a quarter of these went beyond Ziehl-Neelsen staining (as reported elsewhere) [17]. A further limitation of our review is the exclusion of studies that were not available in the English language (n = 20). Conclusion This paper shows the limited quality and applicability of the evidence for the diagnosis and treatment of IGM, which is particularly true for European nations. Whether this truly reflects low prevalence, guidelines to diagnose and manage a rare condition are required, even if they - by necessity - must be based on expert opinion until such a time when robust studies are available. The mobility of the global population further adds to the need for guidelines. The UK serves as a good example: in recent years, as many as 170,000 Hongkongers have applied for British National Overseas status [18], and our review has demonstrated that Hong Kong was the joint-6th most prolific state for publishing on IGM (five publications). Additionally, Iranian nationals were the second largest migrant groups crossing the English Channel in 2024 [19], and Iran was the fourth most prolific state as regards primary research (eleven publications). Thus, if ethnicity is an independent risk factor for IGM, the evolving demographics of the United Kingdom further highlight the need clear guidelines to support UK clinicians. A path to producing clear guidelines in the absence of strong evidence is required and this could be delivered using a Delphi process that establishes a standard diagnostic pathway and management algorithm through consensus. National registries could accompany this, to form a foundation for future research and trials, and to aid the refinement and auditing of produced guidelines. To this end, the Granulomatous Mastitis Registry Study (GRAMAREG) is an ongoing European non-interventional observational study characterising IGM symptoms, their duration depending on treatment, and secondary outcomes such as imaging findings and recurrence [20]. Without such efforts, delayed diagnosis and poor outcomes may remain the norm. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability statement Data are available upon reasonable request. Ethical approval Ethical approval not required. CRediT authorship contribution statement Daniel Ahari: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Writing – original draft, Writing – review & editing. Leah Argus: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Writing – original draft, Writing – review & editing. Rute Castelhano: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. Shaneel Shah: Conceptualization, Data curation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing. Goonj Johri: Conceptualization, Data curation, Investigation, Methodology, Supervision, Writing – original draft, Writing – review & editing. Emma MacInnes: Conceptualization, Formal analysis, Methodology, Supervision, Writing – original draft, Writing – review & editing. Christopher Darlow: Conceptualization, Supervision, Writing – review & editing. Iain Lyburn: Conceptualization, Supervision, Writing – review & editing. Nisha Sharma: Conceptualization, Supervision, Writing – review & editing. Vijay Sharma: Conceptualization, Supervision, Writing – review & editing. Rudresh Shukla: Conceptualization, Supervision, Writing – review & editing. Roisin Bradley: Conceptualization, Supervision, Writing – review & editing. Nazina Arafin: Methodology, Project administration, Supervision, Writing – review & editing. Cliona C. Kirwan: Conceptualization, Formal analysis, Investigation, Methodology, Project administration, Supervision, Writing – original draft, Writing – review & editing. Acknowledgements The ENIGMA Delphi consensus meeting was supported by the Association of Breast Surgeons (UK) Research Development Grant to the value of £9600. C. C. K. is supported by a Masonic Charitable Foundation Fellowship (Royal College of Surgeons/Masonic Charitable Foundation Chair of Surgical Trials grant of £500, 000). I. L. is partially supported by Cobalt Medical Charity, Cheltenham. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.breast.2026.104825. Supplementary data The following is the Supplementary data to this article: Multimedia component 1 References Associated Data Supplementary Materials Multimedia component 1 Data Availability Statement Data are available upon reasonable request.