Lost in Translation: Why Biologic Therapies for Intervertebral Disc Degeneration and Low Back Pain Have Not Reached the Clinic (Yet)

Abstract

Low back pain (LBP) is the leading cause of disability worldwide, imposing substantial social and economic burdens and affecting more than 600 million individuals globally. According to the Global Burden of Disease 2021 study [1], the prevalence of LBP is projected to increase by 36.4% during the next 25 years, with the most pronounced rises expected in Asia and Africa, largely driven by population growth. This anticipated surge, together with progressive population aging, will further amplify its socioeconomic impact, which is already comparable to that of cardiovascular diseases, cancer, and major mental health disorders [2].

Although most episodes of LBP are classified as “nonspecific”, intervertebral disc degeneration (IDD) is frequently implicated in discogenic LBP and considered a major contributing factor. In the context of IDD, the intervertebral disc (IVD) undergoes progressive cell loss and senescence, extracellular matrix (ECM) degradation with dehydration and disc height reduction, as well as pronounced neoangiogenesis and neoinnervation, immune cell infiltration, and deterioration of native biomechanical properties [3]. Despite these structural alterations, a substantial proportion of individuals with disc abnormalities remain asymptomatic [4]. While most pain flares resolve spontaneously in symptomatic cohorts, approximately 5%–10% of patients develop chronic LBP, the management of which remains challenging, as no conservative or surgical strategies directly target the underlying pathophysiological processes [3].

Over the past decades, several biologic therapies have been investigated in both preclinical and clinical settings, with the putative aim of regenerating the degenerative IVDs or, at minimum, inducing local reparative processes presumingly associated with alleviation of discogenic LBP. These approaches include cell-based therapies (particularly mesenchymal stromal cells [MSCs]), platelet-rich plasma (PRP), extracellular vesicles (EVs), biomaterials, growth factors, gene therapy, as well as their various combinations [3, 5, 6]. Although many of these strategies have demonstrated encouraging “regenerative” outcomes, translation from preclinical investigations to clinical trials has been inconsistent and limited to a small number of products. Nonetheless, despite a generally favorable safety profile, the application of these biologic therapies at the bedside has generally yielded only modest improvements in pain and function, with no consistent demonstration of structural repair on imaging [7]. At present, the disconnect between preclinical promise and clinical reality is tempering enthusiasm for biologics in spine care and underscores the need for a critical appraisal of the factors underlying the stalled translational pipeline.

Rather than revisiting the current literature on biologics (which has been extensively reviewed elsewhere [3, 5-7]), this study focuses on the determinants of their limited clinical translation. In particular, we examine the potential roles of patient selection, mechanistic targeting, and alignment between biological effects and clinical endpoints. Through systematic mapping of the biologic therapy landscape and thematic analysis of key failure drivers, we aim to provide an evidence-based assessment of current translational barriers. Building on this analysis, we propose a failure-informed roadmap that emphasizes refined patient stratification, improved concordance among mechanisms, models, and endpoints, optimization of clinical trial design, and clearer definition of the strategic role of biologics within spine care.

This review is therefore positioned not as further advocacy for biologic therapies, but as a critical evaluation of the factors that have thus far limited their clinical impact, with the aim of fostering more rigorous, evidence-based research and advancing precision medicine–oriented approaches for LBP.

Despite extensive preclinical research on biologic strategies for IDD, important uncertainties remain regarding the mechanisms these therapies aim to target. A central issue is the frequent conflation of repair and regeneration. Many biologic interventions are described as regenerative, yet their effects often resemble partial repair or mitigation of tissue damage rather than true restoration of disc structure and function. This distinction is critical because the adult IVD has limited intrinsic regenerative capacity, particularly at advanced stages of IDD.

For example, the formation of cartilage-like tissue within the IVD does not necessarily restore the specialized NP matrix, which contains substantially higher proteoglycan-to–type II collagen ratios than articular cartilage [50, 51]. Although MSCs can differentiate toward a chondrogenic phenotype, this does not ensure production of the unique ECM required to restore native tissue architecture and biomechanical function. Such approaches may nevertheless provide clinical benefit by slowing IDD progression or replacing the degenerative tissue with a less catabolic or less nociceptive phenotype [52]. However, these outcomes are more accurately described as repair or functional replacement rather than true regeneration.

Another unresolved question is which structural alterations within the IVD remain biologically reversible. While early biochemical changes in IDD may retain some degree of plasticity, later structural abnormalities such as annular fissures or cartilaginous endplate (CEP) sclerosis likely represent largely irreversible pathologies. During normal development and maturation, increasing mechanical demands are accommodated through coordinated alignment of cells and ECM components regulated by biochemical and mechanical signals [53]. Degeneration disrupts this organization and alters the native mechanical environment, generating aberrant cues that may impair cellular responses and tissue remodeling [54]. Whether biologic therapies can overcome these chronic mechanical disturbances remains uncertain. In parallel, aging and IDD are associated with increased cellular senescence and apoptosis, resulting in both reduced cell numbers and a higher proportion of metabolically unresponsive cells [53, 55-57]. As a consequence, biologic agents that depend on resident cell responsiveness, such as growth factors, PRP, or EV-based therapies, may have limited efficacy in advanced IDD due to the reduced availability of viable target cells.

The nutritional microenvironment of the IVD represents an additional constraint [58-60]. As the largest avascular tissue in the body, the disc relies on diffusion through the CEPs to supply oxygen and glucose while removing metabolic waste. IDD compromises this transport system through CEP calcification, reduced CEP permeability, and declining proteoglycan content, collectively limiting nutrient diffusion and promoting an acidic microenvironment [55, 61, 62]. Computational and patient-specific diffusion modeling suggest that nutrient availability in many degenerated IVDs may already approach the threshold required to sustain cell viability. Under these conditions, therapies that increase cell density or metabolic activity may further elevate demand within an already constrained niche [63-65]. While the degenerative IVD reportedly contains approximately 1–2 million cells/mL, the number of cells delivered in clinical studies varies widely, ranging from near-physiological levels to more than 60-fold higher concentrations [7, 47]. Such variability raises critical questions regarding cell survival in the nutrient-limited IVD environment and the potential biological consequences of widespread cell death, including the release of intracellular byproducts following starvation or apoptosis. Despite these concerns, these aspects remain poorly characterized in the available literature.

Timing may represent another critical yet scarcely defined variable. Clinically, patients with chronic LBP rarely present during early degenerative stages, meaning structural damage is often already established at the time of treatment. A therapeutic window likely exists in which biologic interventions can meaningfully influence disease progression. Early stages, characterized by viable cell populations and partially preserved ECM architecture, may remain responsive to biologic modulation, whereas advanced IDD marked by fibrosis, structural collapse, and mechanical instability likely exceeds the reparative capacity of such approaches. Consequently, the mechanistic basis of many biologic therapies for IDD remains incompletely understood [66], highlighting the need to better define the limits of tissue reversibility, the influence of the disc microenvironment, and the relationship between structural modification and clinical outcomes. These mechanistic uncertainties complicate the rational design of biologic therapies and may partly explain why promising preclinical findings have not consistently translated into clinical success.

A central limitation of IDD preclinical research is species-specific disc biology, which can profoundly shape degeneration kinetics, repair capacity, and the success of investigational therapies. Many widely used small animal models, like mice or rats, offer established experimental control and behavioral testing options, yet typically retain notochordal cells into adulthood and differ from humans in IVD size, biomechanics, and transport properties, factors that may limit direct extrapolation to adult human IDD [16, 67, 68]. Rabbits, commonly used for puncture-induced IDD because of procedural accessibility and disc size, likewise retain notochordal cells and may exhibit relatively robust intrinsic healing, making them valuable for early-stage assessment, but are likely too optimistic for translation [69]. In contrast, large-animal models (e.g., sheep, goat) better approximate human IVD dimensions, CEP morphology, and mechanical environment, supporting surgical reproducibility and implant or biomaterial testing, albeit with greater cost, longer time courses, and higher regulatory burden [70, 71]. Likewise, porcine models can provide relevant loading environments and biochemical similarities, yet notochordal cell retention may persist depending on breed, and practical constraints often confine their use to later-stage validation [16, 23, 72]. Importantly, canine models illustrate how species selection can either hinder or enhance clinical relevance: non-chondrodystrophic breeds maintain a notochordal-rich, gelatinous NP longer and are less representative of adult human IDD [73], whereas chondrodystrophic breeds develop more human-like, age-associated IDD and are increasingly viewed as a clinically relevant intermediate-sized model, particularly for questions involving pain-related outcomes [74, 75]. Nevertheless, small animal models remain widely used due to cost efficiency, regulatory expectations, and the requirement for efficacy data in these systems before progression to large animal studies, despite their recognized limitations.

Even when degeneration is successfully induced, many models reproduce IDD without faithfully modeling discogenic LBP, not differently from their human counterpart. In discogenic pain research specifically, most studies are performed in rodents, supported by established behavioral testing batteries; however, their translational potential to humans remains limited, and conventional imaging, although a gold standard for structural change, cannot reliably identify disc-related pain [43, 76].

A related challenge is that widely used induction paradigms, particularly puncture-based injury models, often create an acute, externally applied (“outside-in”) insult that differs from the chronic, spontaneous (“inside-out”) trajectory often implicated clinically. Such models may disproportionately emphasize the role of the AF and NP while underrepresenting contributions from CEPs, vertebral bone, and posterior elements, tissues that can be relevant to human IDD and pain biology. Therefore, they should be better regarded as models of acute injury rather than chronic degeneration. Methodological heterogeneity (e.g., needle gauge, depth, dwell time, number and site of punctures, among others) further complicates cross-study comparison and reproducibility [23].

Finally, many preclinical studies foreground short-term outcomes and surrogate readouts (e.g., histology scores, MRI, or biomarker panels) while providing limited evidence for durability and functional relevance. Structural improvement does not necessarily imply restoration of disc mechanical performance, stabilization of the motion segment, or sustained modulation of nociceptive signaling [16]. Recent work emphasizes that behavioral assays are informative but can be confounded by stress, anxiety, and environmental conditions, reinforcing the need for integrated outcome frameworks that combine structure, biology, biomechanics, and pain-related function over clinically meaningful timeframes [23, 67]. Collectively, these limitations argue for a stricter alignment among model choice, disease mechanism, and endpoint selection, and for prioritizing disease models and study designs that better capture human-relevant disc biology, pain-relevant pathways, and durable functional outcomes rather than short-term structural surrogates alone.

As with other investigational medical products, biologic therapies targeting IDD are expected to progress through the conventional translational pathway and demonstrate safety and efficacy across Phase 1–3 clinical trials before being adopted into routine clinical practice. Yet the likelihood of successful translation is notoriously low. It is estimated that approximately 95% of drug candidates entering human trials ultimately fail, with nearly half of these failures occurring during already advanced Phase 3 studies. Consequently, only a very small proportion of experimental therapies successfully translate from bench to bedside, approximately 0.1% [77]. Notably, while the vast majority of biologic therapies investigated in vitro and in preclinical models of IDD have never reached clinical testing, many of the products that have entered clinical practice have been evaluated primarily in non-comparative cohort studies rather than in rigorous randomized controlled trials (RCTs).

To date, MSCs represent the biologic therapy supported by the largest body of clinical evidence, with 5 RCTs published to date (including Phase 2a, Phase 2b, and Phase 3 studies) [78-82]. Early uncontrolled reports consistently described substantial improvements in LBP severity following treatment, occasionally accompanied by modest structural changes on imaging. However, these encouraging results were markedly attenuated in randomized controlled settings. In several RCTs, the differences between MSC-treated and sham-treated groups were small, and although statistical significance was occasionally achieved, the observed improvements frequently failed to surpass the minimal clinically important difference (MCID) for key outcome measures [6]. Regarding other cell types, the recently published Phase 1/2 DiscGenics trial did not show any significant differences between the treatment (high- and low-dose NP progenitor cells) and control groups (vehicle only and saline) [83]. Likewise, the few RCTs investigating the efficacy of intradiscal PRP were characterized by low sample sizes and often demonstrated no significant differences compared to control groups [84-86]. Two NP graft strategies, NOVOCART Disc and VIA Disc NP, have also been evaluated in RCTs. NOVOCART Disc consists of a two-stage procedure in which autologous NP cells are first harvested during discectomy for lumbar disc herniation, expanded, and subsequently injected upon resuspension in a matrix composed of human albumin, hyaluronic acid, and a cross-linking agent. Although treatment resulted in modest reductions of pain scores, these improvements were transient and not accompanied by meaningful structural changes on imaging. Consequently, clinical development of this approach was ultimately discontinued [52, 87, 88]. Conversely, the VIA Disc NP, an injectable human NP allograft, was evaluated in an RCT compared with saline. Both groups demonstrated clinically meaningful improvements in Visual Analogue Scale (VAS) and Oswestry Disability Index (ODI) scores, with a significantly higher responder rate (and number of adverse events) in the VIA Disc NP group [89].

Despite the scarcity of published high-level clinical evidence, international trial registries such as ClinicalTrials.gov list dozens of protocols investigating biologic therapies for IDD. In a previous analysis from our group, we reported that only 26.9% of registered clinical trials had ultimately resulted in a peer-reviewed publication [90]. Despite the recent publication of four additional RCTs [79-81, 83], the proportion of trials failing to reach completion or publication remains substantial. From a clinical trial design perspective, this may reflect several factors.

First, neglecting the inconsistent association between LBP and IDD in most clinical trials creates a substantial risk of selection bias. Many studies enroll patients with imaging evidence of IDD without definitively establishing the discogenic origin of pain, thereby potentially including individuals with non-discogenic LBP and incidental IDD findings. As a result, the likelihood of enrolling patients whose symptoms arise from alternative pain generators despite concomitant IDD is considerable, introducing significant confounding. In this context, the administration of intradiscal biologic therapies may lead to biased estimates of treatment efficacy, as the targeted tissue pathology may not represent the true source of pain.

Second, difficulties in patient recruitment and high attrition rates may further compromise study feasibility and statistical validity, possibly leading to early termination or critical lack of statistical power. Among published trials, loss to follow-up rates range from 9% [52] to 34% [79], posing a substantial threat to the internal validity of reported outcomes. Beyond the burden imposed on participants (often including repeated imaging examinations, blood sampling, and lengthy questionnaire completion), trial attrition may also be influenced by placebo and nocebo effects [91]. These phenomena are particularly relevant in populations with chronic LBP, where, as mentioned, psychological distress and anxiety are common, especially in patients with nociplastic pain features [46].

With regard to study endpoints, most trials rely primarily on patient-reported outcome measures (PROMs), including pain intensity (e.g., VAS), disability (e.g., ODI), and quality of life (e.g., EuroQol [EQ]-5D and Short Form [SF]-36). Structural outcomes, including disc height and NP hydration on MRI, are frequently assessed, though they are often inadequately reported [6, 7]. PROMs are inherently subjective and susceptible to multiple psychological and contextual influences, but remain necessary to determine treatment success or failure in a real-world scenario [92]. At present, no validated surrogate endpoint of pain exists. Although approaches such as quantitative sensory testing, neuroimaging, or inflammatory biomarkers have been explored, none have demonstrated sufficient validity or reliability to replace PROMs of pain in clinical trials [93]. In the IDD field, imaging parameters have long been proposed as surrogate outcomes, yet clinical trials have reported only sporadic cases of disc height preservation or increased disc hydration [6, 7]. Whether such structural changes can be reliably detected (or even occur) within the relatively short duration of most clinical studies remains uncertain. For example, modeling studies suggest that, even under optimistic assumptions of a linear relationship between injected cell number and glycosaminoglycan (GAG) synthesis, administration of 10 × 10⁶ cells would restore only about 65% of the GAG content of a Pfirrmann grade 2 IVD over approximately ten years when applied to a Pfirrmann grade 3 IVD [63]. Such prolonged timeframes are fundamentally misaligned with the duration of most clinical trials and pose a significant translational constraint, apart from having doubtful clinical relevance [66]. Taken together, these observations illustrate a fundamental paradox: clinical trials in IDD rely on subjective measures to evaluate therapies whose objective structural effects are not necessarily associated with their outcomes.

Another methodological aspect that complicates the interpretation of the results of these clinical trials is the use of composite and responder-based endpoints. In addition to (or instead of) reporting mean changes and standard deviations of pain or disability scores, several studies present outcomes as the categories of participants reporting a certain score range, the proportion of patients achieving predefined thresholds of improvement, or the combination of multiple outcomes into composite measures [78-80, 89]. While these approaches may increase statistical power and provide clinically interpretable categories of therapeutic response, they may also complicate the interpretation of treatment effects, particularly when the overall mean differences between treatment groups are small [94]. Methodological studies have highlighted that composite endpoints may amplify the apparent magnitude of benefit when driven by less clinically meaningful components or when thresholds are selected post hoc [95, 96]. In conditions such as chronic LBP, where placebo responses are substantial and treatment effects are often modest, such analytical strategies may contribute to apparently positive findings despite limited absolute differences between intervention and control groups.

Taken together, the most critical limitation is inadequate patient selection, as enrolling heterogeneous populations without confirming a dominant discogenic pain source likely dilutes mechanism-specific treatment effects. This is compounded by reliance on subjective PROMs without validated objective correlates, particularly with the strong placebo responses, and by the mismatch between slow biological repair and short trial durations. In addition, incomplete reporting, including the use of composite endpoints and limited structural outcome data, further limits interpretability. Addressing these factors collectively is likely to yield the greatest improvements in trial validity and near-term translational potential.

Beyond biological and clinical uncertainties, the translation of disc biologic therapies is also constrained by substantial practical, regulatory, and economic barriers. Even when promising signals emerge from preclinical studies or early-phase clinical trials, these challenges can significantly delay or prevent broader clinical adoption [90-97].

Manufacturing complexity represents one of the most significant practical obstacles. Many biologic approaches, particularly cell-based or cell-derived therapies, require highly specialized production processes that must meet strict standards for sterility, consistency, and quality control [97-99]. Autologous therapies typically involve multistep workflows including cell harvesting, expansion, characterization, and reimplantation, each introducing logistical challenges, variability, and costs. In addition, the potency of these products may be affected by patient factors, as individuals with chronic LBP are often older and present with comorbidities that may impair the regenerative capacity of harvested cells [100, 101]. On the other hand, allogeneic approaches may offer greater scalability but introduce additional concerns related to immunogenicity, storage conditions, and long-term safety [97].

Regulatory pathways for biologic therapies are also more complex than those for conventional pharmaceuticals or medical devices. Many disc biologic interventions fall within hybrid regulatory categories such as advanced therapy medicinal products or combination products [98, 102]. These classifications require evaluation of both the biologic component and the delivery system, which may include injectable carriers, biomaterial scaffolds, or surgical transplantation techniques. Navigating these frameworks is often time-consuming and costly, particularly when long-term safety monitoring is required, and clinically meaningful endpoints remain difficult to define.

Risk–benefit considerations further complicate regulatory and clinical adoption. Discogenic LBP is not a life-threatening condition, and symptom severity often fluctuates over time. Consequently, regulatory agencies and stakeholders typically require strong evidence of safety and durable clinical benefit before accepting invasive or costly biologic interventions. Even relatively small risks related to immune reactions or unintended tissue responses may be viewed cautiously when established treatments such as rehabilitation, pharmacologic therapy, or surgery remain available. Demonstrating therapeutic efficacy is also challenging, as structural imaging changes do not consistently correlate with symptoms, while patient-reported outcomes remain subjective and highly susceptible to placebo effects [103-105].

Economic considerations add another layer of complexity. As biologic therapies frequently involve high development and manufacturing costs (translating into substantial treatment prices), healthcare systems and payers might be reluctant to reimburse such treatments without convincing evidence of long-term cost-effectiveness. Such a challenge is further amplified by the modest, delayed, or difficult-to-quantify clinical benefits assessed with conventional outcome measures. Furthermore, the absence of clear reimbursement pathways introduces uncertainty regarding commercial viability and may limit investments from industry and private stakeholders.

These considerations suggest that practical, regulatory, and economic barriers are not independent obstacles but rather downstream manifestations of unresolved biological questions and suboptimal trial design. Even when biological plausibility and early clinical promise exist, successful implementation of disc biologic therapies requires coordinated progress across scientific, regulatory, and health economic domains. Despite the large patient population and substantial market potential associated with such therapeutic tools, these challenges continue to limit the translation of regenerative medicine into widely adopted clinical pipelines.

At present, the field of disc regeneration is wandering in the dark, fearsome “valley of death” of clinical translation [77]. While the remarkable efforts of basic scientists have elucidated key aspects of IDD and laid the foundation for early translational attempts, several critical challenges persist. These include the need for more physiologically relevant in vitro systems, improved ex vivo models that accurately replicate the disc microenvironment and biomechanics, and in vivo models that better reflect the human condition.

Concurrently, clinical research must advance toward a more precise characterization of LBP, with the aim of aligning disc biology, clinical presentation, and stage of degeneration. The integration of clinical, imaging, and molecular data, supported by explainable and ethically robust AI frameworks, holds significant promise in this regard. Such approaches may ultimately reveal that biologic therapies for IDD are most effective within carefully selected patient subgroups, rather than broadly applicable across the heterogeneous chronic LBP population. In parallel, clinical trial methodology requires further refinement to account for the complexity of LBP phenotypes, particularly those influenced by psychosocial factors, which may introduce substantial confounding [90]. The adoption of stratified and adaptive trial designs may help mitigate these limitations and improve the detection of true treatment effects while reducing the risk of bias. Equally important is the need to enhance transparency, standardization, and methodological rigor in trial design and reporting. The implementation and widespread adoption of structured reporting frameworks, such as VERTEBRA [171], may facilitate reproducibility and enable more meaningful comparisons across studies.

Despite significant advances, key aspects of disc biology and treatment response remain poorly understood or assessed, underscoring the need for continued rigorous, collaborative research to translate biologic therapies into reliable clinical therapeutics. This decisive step may prompt researchers to reconsider the “regeneration dream”, as the complete restoration of the native IVD phenotype, irrespective of microenvironmental, structural, and neuroinflammatory changes occurring over decades, is likely unattainable, and its clinical relevance remains uncertain. Instead, the field should resignedly shift toward the concept of “repair”, intended as functional stabilization combined with targeted modulation of the biological processes underlying discogenic LBP.

Daisuke Sakai: writing – review and editing, supervision. Luca Ambrosio: conceptualization, investigation, writing – original draft, writing – review and editing, methodology, project administration. Gianluca Vadalà: writing – review and editing, supervision, funding acquisition. Jordy Schol: conceptualization, investigation, writing – original draft, writing – review and editing, methodology, project administration. Clara Ruiz-Fernandez: investigation, writing – original draft, writing – review and editing. Vincenzo Denaro: writing – review and editing, supervision.

This work was supported by NextGenerationEU, NRRP PNC-E3-2022-23683269 PNC-HLS-TA (CUP code: F83C22002880001).

Jordy Schol, Gianluca Vadalà, and Daisuke Sakai are Editorial Board members of JOR Spine and co-authors of this article. To minimize bias, they were excluded from all editorial decision-making related to the acceptance of this article for publication.

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.