Kidney cancer imposed an estimated global burden of more than 430,000 new cases and approximately 156,000 deaths in 2022, and its incidence continues to rise in most regions [1]. Renal cell carcinoma (RCC) comprises a group of histologically and molecularly distinct entities, of which clear cell RCC (ccRCC) is the most common, accounting for roughly 70% to 80% of cases [2]. Although localized ccRCC is frequently cured by surgery or ablation, approximately one third of patients present with synchronous metastases or develop metachronous dissemination after nephrectomy, and metastatic disease remains the dominant cause of kidney-cancer mortality [2][3]. The biological behavior of ccRCC is anchored by biallelic inactivation of the VHL tumor suppressor in the large majority of sporadic tumors, which stabilizes hypoxia-inducible factors, particularly HIF-2α, and rewires transcriptional programs governing angiogenesis, metabolism, and cell motility [4].

Therapy for advanced ccRCC has changed substantially in the past five years. Combinations of immune-checkpoint inhibitors with tyrosine-kinase inhibitors (TKIs) are now first-line standards, and the HIF-2α inhibitor belzutifan has validated the VHL-HIF axis as a direct drug target, first in VHL disease-associated RCC and subsequently in previously treated advanced disease, where it improved progression-free survival over everolimus [3][5][6]. Nevertheless, primary and acquired resistance to these regimens is frequent, and long-term disease control is achieved in only a minority of patients [7]. Hypoxia itself has emerged as a determinant of both response and resistance in metastatic ccRCC, linking the defining molecular lesion of this disease to its clinical intractability [8]. Understanding the downstream machinery that converts hypoxic, VHL-deficient signaling into invasive and metastatic behavior therefore remains a priority.

Matrix metalloproteinases (MMPs) are zinc-dependent endopeptidases that collectively degrade essentially all components of the extracellular matrix (ECM) and many non-matrix substrates [9]. Proteolysis of basement membrane and interstitial matrix is a biochemical prerequisite for tumor cell escape from the primary mass, entry into and exit from the vasculature, and remodeling of metastatic niches. MMPs were implicated in kidney cancer decades ago, and their roles in RCC pathophysiology have been reviewed periodically [10][11]. However, the evidence base has expanded rapidly since 2021, driven by single-cell and spatial profiling of the ccRCC microenvironment, mechanistic studies of MMP regulation, protein-engineering approaches to selective inhibition, and renewed interest in ECM proteolysis as a modifier of anti-angiogenic and immunotherapeutic efficacy. A critical synthesis of this recent literature is lacking.

This narrative review summarizes peer-reviewed evidence published between January 2021 and July 2026 on the roles of MMPs and tissue inhibitors of metalloproteinases (TIMPs) in ccRCC invasion and metastasis. The review is organized by theme rather than by chronology: we first outline MMP biology relevant to cancer invasion, then synthesize expression and prognostic data in ccRCC, then integrate mechanistic and microenvironmental findings, and finally evaluate translational opportunities and their limits, closing with a critical appraisal and outstanding questions for the field. An overview of the MMP/TIMP axis in ccRCC invasion and metastasis discussed in this review is presented in Figure 1.

Figure 1. The matrix metalloproteinase (MMP)/tissue inhibitor of metalloproteinases (TIMP) axis in clear cell renal cell carcinoma (ccRCC) invasion and metastasis. Biallelic VHL inactivation stabilizes hypoxia-inducible factors and establishes a hypoxic transcriptional program in ccRCC cells. Upstream signals, including ERK/MAPK-ETS1/AP-1, metabolic reprogramming (G6PD), nuclear receptors (FXR), and epithelial-mesenchymal transition (EMT) drivers (SPOCK1, CD44, DSS1), converge on induction of MMP-2, MMP-7, MMP-9, and MMP-14, while non-coding RNAs and macrophage-derived exosomes erode TIMP-mediated restraint (TIMP-2, TIMP-3). Gelatinases and membrane-type MMP-14 then degrade basement-membrane and interstitial matrix at the invasive front and at invadopodia, enabling local invasion, angiogenesis, and intravasation. In the circulation, MMP-9 has been associated with circulating tumor cell (CTC) survival (association-based evidence); at distant sites such as bone, tumor and host compartments remodel a metastatic niche. Therapeutic entry points (red) include selective small molecules, engineered TIMPs, anti-MMP-9 antibodies, and rational combinations with VEGF-pathway and immune-checkpoint inhibitors. HIF, hypoxia-inducible factor; ECM, extracellular matrix; TAM, tumor-associated macrophage.

PubMed was searched in July 2026 for articles published between January 2021 and July 2026, using combinations of the terms “matrix metalloproteinase”, individual MMP and TIMP designations, “clear cell renal cell carcinoma”, “renal cell carcinoma”, “invasion”, “metastasis”, “prognosis”, and related variants. English-language original research and review articles were screened by title and abstract, and studies judged directly relevant to ccRCC biology, or to general MMP mechanisms applicable to it, were included; reference lists of included articles were examined for additional recent sources. Because this is a narrative rather than a systematic review, no formal protocol was registered, and the synthesis emphasizes thematic integration over exhaustive enumeration.

Twenty-three human MMPs share a conserved catalytic architecture built around a zinc-binding motif, and they are conventionally grouped by domain organization and substrate preference into collagenases, gelatinases, stromelysins, matrilysins, membrane-type MMPs, and other minor classes [9]. Beyond the catalytic domain, ancillary modules such as hemopexin-like domains and fibronectin-type inserts govern substrate recognition and exosite interactions, features that have become central to the design of selective inhibitors discussed later in this review [9]. All MMPs are synthesized as inactive zymogens whose propeptide shields the catalytic zinc; activation occurs through proteolytic removal of the propeptide or through allosteric and oxidative mechanisms, providing a first layer of regulation that is spatially restricted to the cell surface or the pericellular environment [9][12]. A second layer is provided by the four TIMPs, which bind active MMPs with high affinity and, at balanced stoichiometry, confine proteolysis to defined microdomains [13]. The biological output of the system is therefore determined not by MMP abundance alone but by the local ratio of activated enzyme to inhibitor, a point that has direct consequences for interpreting expression studies in ccRCC tissue.

MMP biology extends well beyond bulk matrix degradation. Gelatinases A and B (MMP-2 and MMP-9) cleave type IV collagen of basement membranes but also process cytokines, growth-factor precursors, and cell-surface receptors, thereby modulating inflammation, angiogenesis, and immune-cell trafficking within tumors [12][13]. Membrane-type 1 MMP (MT1-MMP, MMP-14) occupies a special position in invasion: anchored to the plasma membrane, it concentrates proteolysis at the leading edge of migrating cells, activates pro-MMP-2 in a ternary complex with TIMP-2, and is enriched at invadopodia, the actin-rich protrusions through which cancer cells penetrate matrix barriers [14][15]. Several MMPs also execute intracellular and even nuclear functions, including roles in DNA-damage responses and transcriptional regulation, which complicates the interpretation of both expression data and inhibitor studies [14]. These properties explain why MMP activity in tumors is pro-invasive in most contexts but can be anti-tumorigenic in others, and they foreshadow the difficulties encountered in clinical translation.

Evidence from human tissue indicates that gelatinase dysregulation is a consistent feature of renal carcinoma, but also that abundance and activity must be distinguished. In a cohort of 302 RCC tumors analyzed by immunohistochemistry, expression of MMP-2, MMP-9, and the adhesion receptor CD44 varied with histopathological subtype and was evaluated as a candidate prognostic panel in clear cell and non-clear cell disease [16]. Direct biochemical measurement adds an important qualification: in homogenates of human renal carcinoma, MMP-9 content was higher than that of MMP-2, yet its specific activity was not correspondingly elevated, indicating that a substantial fraction of the enzyme remains latent or inhibitor-bound in tumor tissue [17]. This discordance cautions against equating immunoreactivity or transcript level with proteolytic output and argues for activity-based readouts in future biomarker work. It also has a practical corollary: tissue-level measurements average contributions from malignant, stromal, and immune cells, so localization data from immunohistochemistry and single-cell studies are needed to attribute proteolytic activity to specific compartments [16][17].

Membrane-type MMPs show a similar pattern of selective upregulation. In paired comparisons within human renal carcinoma specimens, MMP-14 exceeded MMP-15 in expression, protein content, and enzymatic activity. This comparison is restricted to the two membrane-type MMPs examined in the cited renal-carcinoma cohort and identifies MMP-14 as the more highly expressed and more active of the two in that setting, without implying dominance over other membrane-anchored proteases [18]. Consistent with a functional role, transcriptomic analyses of clear cell tumors have associated high MMP14 expression with poor prognosis, and MMP-12 has emerged as an independent predictor of postoperative relapse in conventional (clear cell) RCC [19][20]. Beyond tissue, plasma levels of MMP-1, measured together with other circulating proteins, differ between RCC patients and controls and have been proposed as components of minimally invasive diagnostic panels [21]. Germline variation may also contribute: MMP9 genotypes have been associated with RCC susceptibility and clinical features, suggesting that constitutive differences in MMP-9 expression influence disease risk [22].

Large-scale transcriptomic studies have extended these observations across the MMP family. Systematic characterization of MMP family members in kidney clear cell carcinoma datasets identified coordinated dysregulation of multiple family members, several of which correlated with stage, grade, and survival, supporting their evaluation as a biomarker panel rather than as isolated markers [23]. MMP-9 occupies a recurrent position in such analyses: combined assessment of MMP9 and insulin-like growth factor binding protein 1 (IGFBP1) correlated their expression with the immune microenvironment and tumor progression in ccRCC, and an MMP family-based risk model stratified patient outcomes, with serum amyloid A1 (SAA1) identified as an upstream promoter of ccRCC migration acting through an ERK-AP1-MMP axis [24][25]. Efforts to move these signals toward the clinic include a computed-tomography radiomics model that noninvasively predicted MMP9 expression status and carried prognostic information, an approach that could eventually embed MMP biology into routine imaging [26].

Taken together, the expression literature supports four generalizations. First, upregulation of gelatinases and MMP-14 is reproducible across assay platforms and cohorts, but the evidence base is heterogeneous: the direct ccRCC-specific data are mainly transcriptomic and immunohistochemical, whereas the biochemical tissue measurements supporting gelatinase and MMP-14 upregulation were obtained in mixed-histology renal carcinoma cohorts, so the strength of this conclusion differs by molecule and cohort composition. Second, higher MMP expression or content generally tracks with adverse clinicopathological features and shorter survival, although effect sizes and optimal cut-offs vary between studies. Third, several tissue studies analyzed mixed-histology RCC cohorts rather than pure clear cell populations, and their findings should be extrapolated to ccRCC with caution [16]-[18][21][22]. Fourth, the field remains disproportionately reliant on transcript-level inference from public datasets; protein-level and, especially, activity-level measurements are scarce, and the available biochemical data indicate that they cannot be replaced by transcript surrogates [17]. The cellular sources of the measured enzymes, which bulk tissue studies cannot resolve, are considered in the microenvironment section below. Key findings for the individual family members discussed in this and the following sections are summarized in Table 1.

Table 1.Selected matrix metalloproteinases and tissue inhibitors implicated in clear cell renal cell carcinoma invasion and metastasis.

Note: Reference numbers correspond to the numbered reference list of the main text. CTC, circulating tumor cell; EMT, epithelial-mesenchymal transition; RCC, renal cell carcinoma.

Invasion and metastasis unfold within a microenvironment that ccRCC cells share with stromal and immune populations, and recent single-cell and spatial studies have reframed MMP biology as a collective property of this ecosystem rather than of tumor cells alone. The ECM of ccRCC is extensively remodeled and carries prognostic information in its own right, with matrix organization and composition varying across tumor regions [44]. Single-cell transcriptomic atlases of advanced and localized kidney cancer have resolved the cellular sources of matrix-remodeling programs: cancer-associated fibroblasts, tumor-associated macrophages (TAMs), and the malignant cells themselves each contribute distinct repertoires of proteases, inhibitors, and matrix substrates [45][46]. In RCC with venous tumor thrombus, a pattern of locally aggressive spread in which tumor extends into the renal vein and inferior vena cava, fibroblast activation protein (FAP)-positive fibroblasts orchestrate microenvironmental remodeling, linking stromal activation to the distinctive intravascular growth of this disease [47]. Myeloid populations, long recognized as MMP-9 suppliers in many tumors, execute metastasis-promoting functions that include matrix degradation, angiogenesis, and suppression of anti-tumor immunity [13][48]. In this framing, the gelatinase measured in a tumor homogenate originates from several cell types, and therapeutic targeting of stromal MMP sources may be as important as targeting the malignant compartment. This compartmental view also reconciles an apparent paradox in the expression literature: transcript-level signals attributed to “tumor” MMP-9 in bulk datasets may in part reflect infiltrating myeloid cells, which strengthens rather than weakens the case for MMP-9 as a microenvironmental drug target [23][48].

The later steps of the metastatic cascade also bear an MMP signature. A transcriptional metastatic signature derived from single-cell data predicted survival in ccRCC; because this evidence is prognostic and association-based, it indicates that dissemination-related programs are detectable early in primary tumors but does not by itself establish that the signature causally encodes metastatic capacity, which would require functional validation [49]. Circulating tumor cells (CTCs) must survive hemodynamic shear and immune attack in the bloodstream, and comparative transcriptomics identified MMP9 and TYRO protein tyrosine kinase binding protein (TYROBP) as genes associated with CTC survival in ccRCC through adaptation to the tumor immune microenvironment; this evidence is transcriptomic and association-based, and a causal contribution of MMP-9 to CTC survival remains to be validated functionally [50]. At the distant site, single-cell profiling of ccRCC bone metastases revealed a distinct ecosystem in which insulin receptor (INSR)-positive endothelial cells play a pivotal role, underscoring that colonization involves reciprocal remodeling between tumor and host tissue [51]. Clinical detection of these steps is advancing in parallel: systematic evaluation of CTC detection methods in RCC and broader liquid-biopsy approaches now offer platforms through which MMP-related readouts could be captured longitudinally during therapy [52][53]. Hypoxia, the defining feature of VHL-deficient ccRCC, permeates each of these steps and has been shown to shape therapeutic response and resistance in metastatic disease, providing a mechanistic thread that connects the primary lesion to the metastatic niche [8].

Angiogenesis deserves separate emphasis because it is both a hallmark of ccRCC and a process tightly coupled to MMP activity. MMP-9 releases matrix-bound pro-angiogenic factors and remodels the perivascular matrix during vascular sprouting, and MMPs more generally shape the tumor microenvironment in which vessels form [12][13]. The relationship is not unidirectional: pharmacological inhibition of MMPs was reported to support amoeboid angiogenesis, an alternative, protease-independent mode of vascularization that hampered VEGF-targeted therapies through myosin light chain and ERK signaling [54]. This observation derives from non-renal cancer models, and whether the same compensatory route operates in ccRCC is untested; even so, it carries a caution for a disease treated predominantly with VEGF-pathway inhibitors: MMP blockade could, in principle, relieve one invasive mechanism while unleashing another, and any therapeutic strategy directed at proteolysis must anticipate such escape routes.

This review and the literature it summarizes share several limitations. As a narrative rather than systematic review, the synthesis is subject to selection and interpretation biases, although the search strategy was specified and the inclusion window was restricted to recent, peer-reviewed work. Within the primary literature, transcript-level analyses of public datasets predominate, and the biochemical data caution that transcript abundance is an imperfect surrogate for proteolytic activity [17][23]. Mechanistic studies rely heavily on cell lines and xenografts whose microenvironments differ from human ccRCC, and almost no interventional study of an MMP-directed agent has been performed specifically in this disease. In coverage, this review does not attempt exhaustive treatment of all 23 family members: MMPs for which recent ccRCC-specific evidence is scarce, including stromelysins and membrane-type MMPs other than MMP-14, are not discussed individually, and their roles in this disease remain open questions. Finally, publication bias toward positive mechanistic findings is likely, and contradictory or null results are underrepresented in the cited evidence.

Evidence published between 2021 and 2026 positions matrix metalloproteinases as central executors of invasion and metastasis in clear cell renal cell carcinoma. Gelatinases and MMP-14 are consistently upregulated across ccRCC transcriptomic and immunohistochemical studies, with corroborating biochemical evidence from mixed-histology renal carcinoma cohorts that should be extrapolated to pure clear cell disease with caution; their induction is driven by convergent ERK/MAPK-ETS1/AP-1 signaling, metabolic reprogramming, and EMT programs; and their activity is further unleashed by TIMP-3 loss and TIMP-1 overexpression, all within a microenvironment whose stromal and immune compartments both supply and respond to proteolysis. The translational opportunity is real but unrealized: MMP-related readouts are promising components of prognostic and imaging biomarkers, yet none has been prospectively validated for clinical use, and no selective MMP-directed agent has been tested in ccRCC despite a mature protein-engineering toolkit. The field’s priorities are correspondingly clear. Activity-based and spatially resolved measurements should replace transcript surrogates in biomarker development; selective inhibitors should be evaluated in biomarker-selected ccRCC cohorts with rational combinations against VEGF and immune-checkpoint pathways; and trial designs must respect the context-dependent, occasionally protective functions of this protease family. If these conditions are met, the MMP axis could move from a descriptive marker of aggressive disease to an actionable component of ccRCC management; until then, its direct impact on the standard of care lies in refining prognostic stratification and in identifying the patient subsets most likely to benefit from future protease-directed trials.

During the preparation of this work, the authors did not use any artificial-intelligence-assisted writing tools. All content was written, reviewed, and edited solely by the authors, who take full responsibility for the integrity and accuracy of the manuscript.

B. L. conceived the review, performed the literature search, and drafted the manuscript. B. L. curated and verified the references. Y. L. supervised the work and critically revised the manuscript. All authors read and approved the final version.