Expert Perspective: An Approach to Refractory Lupus Nephritis

Clinical Challenge

Case:

A 29 year-old woman with a 5-year history of systemic lupus erythematous (SLE), manifested by arthralgia, skin rash, positive anti-double stranded DNA antibodies and low complement components 3 and 4 (C3, C4), on low-dose prednisone and hydroxychloroquine, presented with new onset nephrotic range proteinuria (3.5g/day). Her serum creatinine (SCr) concentration was 0.6mg/dL. A kidney biopsy showed Class IV lupus nephritis (LN) with wire loop lesions, endocapillary hypercellularity, mesangial hypercellularity, 1 glomerulus with karyorrhectic debris and mild interstitial fibrosis (Fig. 1). National Institutes of Health activity and chronicity indices were 7/24 and 1/12, respectively. She was initially treated with three intravenous doses of methylprednisolone totaling 2g followed by prednisone, 0.5 mg/kg/d with a tapering schedule, and intravenous cyclophosphamide, 500 mg every 2 weeks for 6 doses. Cyclophosphamide was to be followed by mycophenolate mofetil (MMF) maintenance (2g/d), but at her 12 week follow-up proteinuria was 3.4 g/day and SCr was 0.8mg/dL. Given this lack of response, MMF was started at 3g/d, tacrolimus 1 mg bid was added, prednisone was recycled to 0.5 mg/kg/d and the taper restarted. Despite these medication changes, the patient had persistent proteinuria (3 g/d) and hypocomplementemia 6 months later. The patient was given two 1g doses of rituximab and MMF was continued. Some progress was noted with improved complement levels, but the patient became leukopenic and MMF had to be reduced to 500 mg/d. Voclosporin became available around this time and was added. The patient’s proteinuria decreased to 500 mg/d.

Figure 1: The kidney biopsy of a patient with refractory lupus nephritis.

Light microscopy demonstrating (A) wire loop lesions and (B) mesangial and endocapillary hypercellularity. Electron micrographs showing (C) mesangial and D) subendothelial immune complex deposits.

Background

Systemic Lupus Erythematosus (SLE) affects the kidneys in about half of all patients (1). Lupus Nephritis (LN), the most common manifestation of kidney involvement in lupus, typically occurs within the first 6–36 months of diagnosis, and in 25–50% of patients LN is the initial presentation of SLE (2). Despite prompt diagnosis and treatment with aggressive immunosuppression, it has been reported that 14–33% of LN patients fail to respond and are refractory to treatment (3, 4).

If no response to a single course of one standard therapy is taken as refractory disease, then we suspect that the reported or perceived frequency of refractory LN is largely overestimated. Several factors besides drug resistance contribute to a poor response to treatment and should be excluded before labeling a patient as treatment resistant. For example, non-adherence to prescribed treatment is very common among patients with autoimmune diseases. Using a Medicaid Database, adherence of over 4000 patients with SLE was assessed during their first year of therapy using the criterion of medication refilled >80% of time as “adherent”. Fewer than 25% of patients met this metric (5). Female sex, younger age, Black race and Hispanic ethnicity were associated with higher odds of non-compliance. Other important barriers identified were health literacy, financial issues, access to healthcare, perceived treatment inefficacy and side effects (6). It is therefore prudent to assess adherence before changing treatment plans or adding additional immunosuppression. Obtaining pharmacy refill records, doing pill counts and monitoring drug levels may help in ascertaining adherence. Monitoring hydroxychloroquine levels also does seem to improve adherence (7). Hydroxychloroquine has a long, 40–50 day half-life, and serum levels do not fluctuate much, so an undetectable hydroxychloroquine level suggests long-term non-adherence (8). Once non-adherence is unmasked a non-judgmental conference with the patient (and their family) should be pursued, incorporating open discussion to identify and address barriers, improve disease understanding, and if possible, simplify the treatment regimen and decrease pill burden (9).

Although standard-of-care (SOC) regimens for LN are well-established, adequate drug dosing can be challenging. Most patients start treatment within prescribed dose ranges, but how an individual metabolizes a specific drug is generally not known a priori. Additionally, dosing is often decreased in response to reported side effects, and this may affect efficacy. Ideally, treatment adequacy for individual patients would be determined by monitoring drug blood levels, but the therapeutic range for most SOC LN drugs has not been established (10). Furthermore, therapeutic drug monitoring (TDM) can be challenging in the clinical setting. Measurement of the area under the concentration-time curve (AUC) is the most accurate determination of an individual’s exposure to a drug, but involves obtaining blood samples repeatedly over time. Trough concentrations are more convenient, and sometimes do correlate with a drug’s AUC (11). MMF dose adjustment could be considered in non-responsive patients if pre-dose levels are consistently <3–4.5 mg/L (10, 11). Therapeutic drug monitoring of the calcineurin inhibitors (CNIs) cyclosporine and tacrolimus is more helpful for monitoring adherence and toxicity than efficacy (10). Voclosporin levels do not need to be monitored (12), and most clinical laboratories do not offer testing. Pharmacodynamic effects based on a drug’s specific mechanism of action may be used in lieu of blood levels. For example, the efficacy of rituximab (RTX) in LN seems to depend on the achieving complete peripheral B cell depletion (13).

Patients with LN must be given sufficient time to achieve a kidney response. The European League Against Rheumatism (EULAR) guidelines point out that patients with nephrotic-range proteinuria at baseline may need an additional six to twelve months to achieve complete clinical response (14). On the other hand, continuing an ineffective therapy for many months waiting for a complete response will likely result in chronic damage to the kidneys. Determining the balance between sufficient duration and prolonging exposure of the kidneys to inflammatory injury may be guided by the observation that early, albeit partial response to treatment is associated with complete response later. A decline in proteinuria of at least 25% at 8 weeks has been shown to associate with a 50% reduction in proteinuria after 6 months of therapy (15), and a fall in proteinuria of at least 50% after 6 months of therapy predicts good long-term kidney survival (16). Patients who have a reduction in proteinuria of ≥25% after 2–3 months of treatment are expected to continue to show improvement and may not need to switch therapy; on the other hand patients who do not show improvement are candidates for a change in treatment. It is important to emphasize that these thresholds are only intended to provide a starting point for decision making and are not rigid. Some patients who did not meet these thresholds went on to have good long-term responses, and conversely, some patients who did achieve these metrics did not do well (15, 16). Monitoring LN should be a continuous process using real-time data for flexible decision making.

While there is no consensus definition for a complete kidney response, most clinical trials require proteinuria to fall to ≤700–500 mg/d and SCr or estimated glomerular filtration rate (eGFR) to be within 10–25% of baseline (15,17). Ideally, a complete kidney response based on these clinical thresholds would reflect resolution of intrarenal inflammation, but there is considerable discordance between clinical and histologic responses (18–20). Proteinuria and SCr reflect both acute inflammatory kidney injury and chronic kidney damage. Therefore persistent proteinuria and/or an elevated SCr must be correctly interpreted. This is particularly difficult for LN patients who have had their disease for a long time, or who have experienced multiple LN flares, and in whom accumulated chronic damage may result in proteinuria and kidney function that will never meet complete clinical response criteria. Persistence of proteinuria or an abnormal eGFR in patients who have been treated for more than 18–24 months therefore does not always represent ongoing immune-mediated kidney injury that has evaded treatment (21). A kidney biopsy may be needed to differentiate active disease from chronic damage and resolved inflammation in such cases (18, 19). A kidney biopsy may also demonstrate an unanticipated process, such as antiphospholipid nephropathy, that would require a different intervention.

Finally, kidney disease in LN patients may progressively worsen despite adequate therapy because of genetic factors, giving the appearance of non-responsiveness. For example, the genetic variations of apolipoprotein L1 (ApoL1) that predispose patients of African ancestry to end-stage kidney disease (ESKD) (22) are found in patients with LN and are associated with ESKD and a shorter timeline to ESKD (23). Progressive kidney failure in such individuals may not be due to treatment resistance (24). Alternatively, some genetic conditions, like auto-inflammatory disorders characterized by increased type 1 interferon production, may have similar clinical and pathologic findings as LN, but do not respond well to conventional LN treatments (25).

Approach

Once a LN patient is diagnosed as treatment-resistant they often are exposed to more and more potent immunosuppressive treatments, sequentially or in parallel (26, 27), and are at ever increasing risk of serious short and long-term adverse events. After SOC treatments are exhausted, the evidence of efficacy for any of the proposed salvage therapies is very limited. Given the paucity of evidence for alternative therapies, and keeping in mind the previously discussed situations that may mimic treatment-resistance, we consider refractory LN as no response (or worsening) of proteinuria and/or eGFR to two different SOC induction regimens after 4–6 months in patients who are taking their prescribed medications in doses that are generally believed to be therapeutic. This approach is operationalized in Figure 2.

Figure 2:

Approach to the diagnosis of refractory lupus nephritis.

The proposed algorithm takes into account patient adherence, therapeutic drug monitoring, and tries to balance time in treatment against accumulating chronic kidney injury. These attributes are evidence-based, at least to the extent that there is evidence. However, the proposal to sequentially try two different SOC induction regimens is more opinion-based, and finds rationale in the heterogeneity of LN. Not every patient is expected to respond to the same medication. Over half the patients enrolled in all contemporary clinical trials of LN, even those considered successful, did not achieve a clinical complete response by the end of the trial. It seems reasonable to try another established regimen before reaching for rescue therapies. This approach is consistent with the independent LN guidelines from the EULAR, American College of Rheumatology (ACR) and Kidney Disease: Improving Global Outcomes (KDIGO) consortium (14, 28).

The LN guidelines suggest adherent patients who are first treated with MMF + glucocorticoids should be switched to Euro-lupus or NIH dosing of cyclophosphamide (CYC) + glucocorticoids and vice-versa. We suggest that after poor response to the first therapy used, strong consideration be given to switching to a regimen that adds voclosporin or belimumab, one of the newly approved LN therapies, to SOC. The addition of each of these drugs to background SOC significantly improved the number of patients having a good treatment response (29, 30). Voclosporin responses were better in patients who came into the trial already on MMF, and belimumab showed a larger beneficial effect in relapsed patients compared to de novo LN (31, 32). These observations suggest that both of these novel therapies improve response rates in patients who have already been exposed to some immunosuppression. Of note, belimumab may be less effective in patients with proteinuria ≥3g/d, and voclosporin use should be used cautiously or not at all in patients with a significantly impaired GFR of <45 ml/min (29, 30). If voclosporin is not available, tacrolimus could be considered, assuming a CNI class effect (33, 34).

Given that belimumab and voclosporin are new tools for LN management, there may be an inclination to use these medications as rescue therapies for refractory LN. However neither drug has been systematically evaluated in refractory patients, and such patients were excluded from the pivotal trials of voclosporin and belimumab. Interestingly, there are reports that the addition of a CNI (tacrolimus or cyclosporine) to MMF may result in a response in refractory or relapsing LN, but these have been small uncontrolled studies (33, 35).

Once a diagnosis of treatment-resistant LN has been established, the most common next step has been the addition of B-cell targeted therapy, specifically RTX. A fair amount of data on the response of refractory patients to RTX has been published, and while generally of poor to modest quality, provide some evidence for its use. Evidence for any particular therapy in the case of RTX failure is severely limited. The risk-benefit of increasing immunosuppression in these patients must be assessed. In such cases it is prudent to assess the status of the patient’s immune system and bone marrow by measuring quantitative immunoglobulin levels and leukocyte subset levels (36). If uncontrolled LN has been going on for a while, it is often useful to do a kidney biopsy and assess chronic damage. Patients with severe interstitial fibrosis, tubular atrophy and global glomerulosclerosis have irreversible kidney damage and will inevitably need dialysis. The amount of viable kidney parenchyma left to save and the likelihood of need for kidney replacement therapy even if this parenchyma is saved will figure importantly in calculating the risk to benefit of intensifying immunosuppression further.

Of note, although race and ethnicity may associate with an increased likelihood of developing refractory LN, once a patient is declared refractory, as defined above, we do not choose subsequent rescue therapies based on race or ethnicity. Treatment options are too few and evidence for differential efficacy by race or ethnicity too sparse to make this feasible.

Our patient was compliant with therapy. Despite treatment with Eurolupus CYC, MMF, and tacrolimus the patient did not respond. Our approach to rescue therapy for patients such as this is shown in Figure 3. The evidence supporting this approach is discussed below.

Figure 3:

Therapeutic approach to the patient with refractory lupus nephritis.

Evidence

Rituximab

In the development of LN, B cells play a central role in producing pathogenic autoantibodies, initiating release of cytokines like IL-6 and TNF alpha and activating T cells by providing co-stimulatory support. B-cell directed biologics deplete or impair the function of B cells. Although rituximab (RTX) does not deplete plasma cells directly, it prevents repletion of plasma cells by depleting precursor B cells (37).

RTX, which is directed against the B-cell surface molecule CD20, is a chimeric mouse-human monoclonal antibody. Many uncontrolled studies and open-label observational studies have reported efficacy of RTX in patients with refractory LN with response rates of 50% to 80% (38–45). Table 1 summarizes these RTX data. Of note, in these studies there was significant variation in how refractory LN was defined, and relapsing patients were frequently clustered with refractory patients. Concomitant therapies and duration of follow up also varied considerably between studies.

Table 1:

Rituximab for refractory lupus nephritis: Selected studies

Reference	N	Refractory or relapsing lupus nephritis	Follow up (months)	Rituximab Dose Regimen	Response to rituximab
Gunnarsson et. al (38)	7	CYC resistant LN	6	375 mg/m2/week × 4	SLEDAI scores improved, decrease in the renal activity index- on repeat biopsy at 6 months
Goswami et.al (39)	14	Refractory or relapsing LN	6	Not specified	CR 71.4% PR 28.6%
Davies et. al (40)	18	All patients had severe, active disease and had failed conventional therapy including MMF and CYC	6	Two 1 g infusions two weeks apart.	CR 61.1% PR 11.1%
Melander et. al (13)	20	Eighteen patients (90%) had already received at least one conventional therapy including intravenous CYC in 15 patients, with a median cumulative dose of 6 g	22	375 mg/m2/week × 4	CR 35% PR 25%
Contis et.al (41)	17	All patients with refractory LN defined as resistant to standard treatment with CYC	12	375 mg/m2/week × 4 (10 patients) or two infusions of 1 g at day 0 and day 15 (seven Patients).	CR or PR 53%
Jonsdottir et. al (42)	25	23 patients refractory to conventional therapy including CYC and/or MMF	12	375 mg/m2/week × 4	CR 64% PR 24%
Lindholm et.al (43)	17	signs of active kidney inflammation despite ongoing treatment with CYC (n = 14) or MMF (n = 3)	6–12	375 mg/m2/week × 4	CR 12% PR 53%
Iwata et. al (44)	63 with SLE, 36 with LN	All refractory to high-dose steroids and various conventional therapies including CYC (54%), MMF (16%), CNI (14%)	12	500 mg twice at one-week intervals (days 1 and 8) in 22 patients; 500 mg at one-week intervals (days 1, 8, 15 and 22) in nine patients; 1000 mg infused at two-week intervals (days 1, 15) in seven patients; 1000 mg infused (days 1, 15, 168, 182) in 25 patients	BILAG score improved in 83.3% UPCR decreased significantly
Iaccrino et.al (45)	145 with SLE, 68 with LN	All had failure of at least one immunosuppressant	12	118 with two infusions (1 g), two weeks apart; 27 with 375 mg/m2/week × 4, followed in 10 cases by two further doses, after 1 and 2 months	CR 30.9% PR 63.2%

Abbreviations: CYC, cyclophosphamide; MMF, mycophenolate mofetil; CNI, calcineurin inhibitor, CR, complete response; PR, partial response; SLEDAI, Systemic Lupus Erythematosus Disease Activity Index

Based on a systematic analysis of 26 studies that described 300 patients with refractory LN, defined as being un-responsive to previous therapy with one or more immunosuppressive agents, the addition of RTX converted 40% of the refractory patients to complete clinical kidney responders, and 34% to partial responders (46). In the reports selected for this systematic review, a RTX dose of 375 mg/m² once a week × 4 was most commonly used (49%), followed by 1000 mg x2, two weeks apart in 37% of patients. Thirty percent of cases received CYC along with RTX, 25% received MMF, 7% received azathioprine and 4% received methotrexate (46). Treatment responses were more common in Class III LN and less frequent in Class IV or V LN. Another meta-analysis of 31 studies described 1112 patients with refractory lupus, from which 10 studies with 223 refractory LN patients showed that 46% and 32% of patients achieved a complete and partial kidney response, respectively after RTX was added. Refractory disease was defined as resistance to traditional therapy; use of prior therapies was not described (47). Further large well-designed multicenter randomized controlled trials are warranted to establish the role of rituximab in refractory LN. In the future, obinutuzumab, an anti-CD20 monoclonal antibody like rituximab but more potent may be evaluated in refractory LN, but at present it is in a phase III trial after a successful phase II trial (48).

Several other approaches to refractory LN treatment have been reported (Table 2). In general, the evidence supporting these therapies is minimal and of low quality. Nonetheless, depending on a patient’s specific situation, it may be necessary to consider these alternatives.

Table 2.

Other therapies tried for refractory lupus nephritis

Reference	Therapy	N	Refractory/Relapsing LN	Follow up (months)	Response
Choi et. al (33)	MMF plus tacrolimus	29	12 Refractory and 17 relapsing	12	CR 25.9% PR 29.6%
Jesus et.al (35)	MMF plus tacrolimus	17	MMF-resistant patients (30)	6	CR 35% PR 35%
Segarra et. al (51)	Bortezomib plus dexamethasone	12	Refractory	9–30	CR 8.3% PR 83.3%
Zhang et.al (52)	Bortezomib plus dexamethasone	6	Refractory	6–24	CR 60% PR 20%
Ostendorf et.al (55)	Daratumumab	1	Refractory	12	Proteinuria improved
Tam et. al (58)	Leflunomide	17	Refractory or intolerant to conventional treatment	12	CR 29% PR 47%
Levy et. al (61)	Intravenous immunoglobulin	7	Refractory, patients who failed CYC	6	Decreased proteinuria in all patients
Monova et.al (63)	Intravenous immunoglobulin	58	Refractory	7 years	CR 30% PR 40%
Zhang et.al (65)	Interleukin-2 therapy	10	Refractory to at least two conventional treatments	6	Decreased proteinuria
Pickering et. al (67)	Eculizumab	1	Refractory to CYC, rituximab, MMF and tacrolimus	18	Decreased proteinuria, improved renal function
Mougiakakos et.al (68)	CAR-T	1	Refractory to CYC, MMF, Tacrolimus, Rituximab, belimumab	1.5	CR achieved

Abbreviations: LN, lupus nephritis; CYC, cyclophosphamide; MMF, mycophenolate mofetil; CR, complete response; PR, partial response

Anti-plasma cell therapy

An underlying pathogenic mediator of refractory LN could be long-lived autoreactive plasma cells that are resistant to commonly used immunosuppressive therapies (49). Bortezomib, a proteasome inhibitor, eliminates plasma cells, thereby reducing autoantibody production, blocks T-cell-dependent inflammatory responses, and decreases interferon-alpha induction by disrupting Toll-like receptor signaling in dendritic cells (50).

A series of 12 patients resistant to induction therapy with CYC, steroids, MMF, and rituximab, were treated with bortezomib plus dexamethasone. A complete clinical kidney response was achieved in one patient and 11 patients had a partial response with improvement in proteinuria, SCr and serological markers after a mean of 6 bortezomib cycles (51). Similarly, a series of five patients with refractory LN also showed reduction of proteinuria and improved kidney function with 4 cycles of bortezomib plus glucocorticoids. Over 6 to 24-months of follow-up, three patients achieved a complete response, one had a partial response, and one patient progressed and required kidney replacement therapy (52). Like other B-cell therapies, bortezomib may cause hypogammaglobinemia requiring intravenous immunoglobulin rescue (51, 53). Also concerning, patients receiving bortezomib may develop a disabling peripheral neuropathy, although the incidence of this adverse event is decreased if subcutaneous dosing is used (54).

Another approach to plasma cell depletion is through the anti-CD38 monoclonal antibody daratumumab, which kills plasma cells and modulates effector T-cell responses (55). A recent case report of two patients with life-threatening, refractory SLE, one of whom had LN showed excellent clinical and serological responses to daratumumab given weekly for 4 weeks followed by longer-term belimumab (55). The LN patient had an improvement in proteinuria from over 6 g/d to around 1 g/d and normalization of SCr levels during the 12-month follow-up period.

While this report, which describes only two patients, provides low-quality evidence within the hierarchy of types of clinical investigation, it offers potential mechanistic insights into refractory LN. Both of these refractory lupus patients had been treated with bortezomib prior to daratumumab, but did not achieve satisfactory responses. This suggests, at least for some patients, that depleting long-lived plasma cells will not be sufficient to control lupus activity. A deeper dive into potential cellular targets for daratumumab showed that in addition to plasma cells, CD38 is expressed on plasma-blasts, mature B cells, and plasmacytoid dendritic cells in lupus patients, and that CD38 expressing T cells are expanded (55). Thus targeting CD38-expressing cells of various types may have contributed to the effect of daratumumab.

An immunoproteasome inhibitor, KZR-616, is current under investigation for LN, but not specifically for refractory (56).

It is important to keep in mind that bortezomib and daratumumab only transiently deplete plasma cells; this effect must be maintained using additional immunosuppression to prevent autoreactive B-cell precursors from developing into autoreactive plasma cells (55). This was the rationale for following up daratumumab with belimumab after lupus came under control (55).

Leflunomide

Leflunomide, an inhibitor of dihydroorotate dehydrogenase, targets lymphocytes and has anti-proliferative and anti-inflammatory actions. A meta-analysis comparing leflunomide to CYC suggested a better safety profile and improved efficacy for leflunomide in LN, but similar effects on the systemic lupus erythematosus disease activity index (SLEDAI) (57). Leflunomide was used to treat 17 refractory or intolerant (to conventional immunosuppression) LN patients (58). Subsequently, 76% of the patients achieved a response (complete in 29%, partial in 47%). More extensive evaluation of leflunomide will be needed before it can be recommended for refractory LN.

Intravenous immunoglobulin

Intravenous immunoglobulin (IVIg) is a biologic therapy comprised of polyclonal antibodies derived from the plasma of a large pool of healthy donors. In addition to being used to treat hypogammaglobulinemia, it has the potential to treat inflammatory diseases, cancer and autoimmune diseases. IVIg tips the balance of activating and inhibitory immune responses by neutralizing auto-antibodies through anti-idiotype binding, up-regulating inhibitory Fc-receptors, and increasing clearance of pathogenic autoantibodies via the reticuloendothelial system (59, 60). Seven patients with biopsy proven class IV or V LN and nephrotic syndrome who had failed therapy with cyclophosphamide and prednisone, showed an improvement in proteinuria with 1 to 6 courses of high-dose IVIg (61). Beneficial effects of IVIg have also been shown in a small Italian cohort (n=12) with refractory SLE, and a Bulgarian cohort (n=58) with treatment refractory chronic glomerulonephritis (62, 63). The main advantage of IVIg is that it is not immunosuppressive, and may therefore be useful in patients who have been over-immunosuppressed and are at risk of infection.

Interleukin-2 therapy

Low-dose interleukin-2 (IL-2) has been used to influence the balance of T cells in SLE patients away from conventional (effector) phenotypes to regulatory phenotypes (64). A small series of 10 patients with refractory or relapsed LN was treated with low-dose, recombinant IL-2 and after 12 weeks, seven patients had a fall in proteinuria of ≥50%, and 2 of these had a complete renal response (65). This was accompanied by a significant expansion of peripheral T regulatory cells (65). IL-2 immunotherapy to restore T-cell regulatory homeostasis in LN may be a novel therapeutic approach for resistant LN, but needs to be tested in larger, randomized-controlled trials.

Anti-complement therapy

The importance of the complement system in the pathogenesis of kidney injury in LN is well-established by experimental models, and is believed to translate to human disease (66). Since refractory LN is characterized by persistent kidney inflammation, anti-complement therapies may be useful in controlling this inflammation, especially if complement-driven. While there are now several complement-targeted therapeutics in various stages of development, eculizumab, a monoclonal antibody that binds to complement component 5 (C5) and prevents formation of C5a and the membrane attack complex (C5b-9) has had the longest clinical exposure. Most of the reports of eculizumab and SLE/LN have been in the context of concomitant thrombotic microangiopathy (TMA). However, a report detailed the successful use of eculizumab for the treatment of a patient with severe LN and no TMA who had failed CYC, MMF, RTX and tacrolimus. During follow-up over 18 months, the patient achieved a sustained and rapid improvement in kidney function and proteinuria (67). While much more evidence will be needed, understanding in whom uncontrolled disease is mediated by complement would facilitate the application of anti-complement therapies in resistant LN.

Chimeric antigen receptor-modified T cell therapy

The principle of CAR-T cell technology is to engineer autologous T cells to express a specific antigen receptor so the modified T cells can recognize and only kill those cells that express the antigen (68). This targeted effect of T-cells is much faster and longer lasting compared to monoclonal antibody therapy. Chimeric antigen receptor (CAR)–modified T cells have mostly been developed to recognize CD19 and other B-cell surface antigens for use in refractory or relapsed B cell malignancies. Harnessing B cell-directed CAR T-cells to treat SLE has garnered attention as a way to deplete autoreactive B cells completely and for a long duration (68).

A recent case report described a 20 year old with refractory SLE/LN treated with CAR-T cells after achieving lymphodepletion with fludarabine and cyclophosphamide. Proteinuria decreased dramatically and the SLEDAI score fell from 16 at baseline to 0 at follow-up (68). Another case report described a 41-year-old with stage IV diffuse large B-cell lymphoma (DLBCL) and a 20-year-history of SLE who was treated with CD19-BCMA compound CAR-T cells with dual targeting of CD19 on B-cells and BCMA (CD269) on plasma cells (69). The patient’s SLE remained stable and DLBCL in remission for over 23 months despite receiving no additional immunosuppressive or chemotherapy.

Anifrolumumab

Anifrolumumab, recently approved by the US Food and Drug Administration for the treatment of SLE, is a fully human, IgG1κ monoclonal antibody to the type 1 interferon receptor subunit 1, and inhibits signaling by all type 1 interferons (70,71). Anifrolumumab is currently being evaluated for LN, and results from a phase II trial will be available later this year (72). While there are no current data regarding the use of anifrolumumab in refractory LN, a transcriptomic analysis of protocol kidney biopsies after induction therapy showed that interferon pathway transcripts remained upregulated in the kidneys of patients who did not respond to therapy, but declined in patients who had a complete clinical response after induction (73). These data raise the possibility that LN patients who do not respond to SOC treatment may have ongoing intra-renal interferon activity that could respond to an interferon antagonist.

Plasma exchange

Like IVIg, plasma exchange (PLEX) can be administered without significantly immunosuppressing patients, if plasma rather than an albumin solution is added back. However, the evidence for PLEX or immunoadsorption in refractory LN is minimal, consisting mostly of single case reports and small observational studies. In contrast, a well-designed randomized controlled trial of PLEX added to SOC therapy in patients with severe (not necessarily refractory) LN did improve clinical outcomes (74).

Hematopoietic stem cell transplantation

Autologous stem cell transplantation temporarily resets the adaptive immune system and depletes autoreactive immunologic memory (75). In two large studies of patients with SLE the probability of 5-year disease-free survival was 50% after autologous stem cell transplantation (76, 77). More recently, 22 patients with refractory/relapsing LN underwent autologous stem cell transplantation. At 72 months median follow-up, 18 patients were in complete clinical remission (78). The relapse rate was 27% and treatment-related mortality was 5%. Given the higher risk of short-term mortality associated with autologous stem cell transplantation, and the risk of recurrence after transplant, we consider this approach only when other options are exhausted.

Mesenchymal stem cell transplantation

Mesenchymal stem cells (MSCs) have immunomodulatory properties and have shown therapeutic benefits when given to patients with autoimmune conditions. In patients with LN and in mouse models of LN, transplantation of MSCs has been shown to suppress autoimmunity and restore kidney function (79). MSC transplantation induces regulatory immune cells and suppresses Th1, Th17, T follicular helper cell, and B-cell responses (79). Although several single-arm studies have shown a therapeutic benefit of MSCs in patients with LN refractory to conventional treatments, when tested in a randomized double blind trial, albeit a small cohort, an additional benefit of MSCs over SOC was not observed (80). Better designed, larger, randomized-controlled trials are needed to evaluate role of MSCs in treatment of LN patients.

Discussion

The management of LN that is truly refractory to SOC is challenging diagnostically and therapeutically. There is no standard definition or specific laboratory test for refractory disease, and because non-response to treatment may have several etiologies, refractory LN is a diagnosis of exclusion. It is critical to understand all of the contributing factors to non-response before labeling a patient as treatment resistant, because the consequence of this diagnosis is generally piling on more immunosuppression and increasing the risk of adverse iatrogenic outcomes. We suggest the most common causes of non-response in the lupus and LN populations are non-adherence to treatment and under-treatment with conventional drugs. While adherence issues can be difficult to address, and titrating SOC therapies is difficult without established therapeutic levels for most drugs, neither requires escalating immunosuppression. For patients who are resistant to SOC treatments, the choice of what should be used next is based on evidence that is of low to modest quality. There have been no randomized clinical trials of rescue therapies for LN. As such, the studies supporting regimens for refractory disease are mainly observational, uncontrolled, and suffer from inclusion of a heterogeneous group of patients previously treated with variable immunosuppression. Of all the therapeutic approaches to refractory disease, treatment with the anti-CD20 biologic rituximab has garnered the most attention. There have been a sufficient number of such studies that several systematic analyses have shown improved outcomes in resistant patients after the addition of rituximab. The fact that there appears to be a favorable signal for anti-CD20 suggests this may be a reasonable first approach to the refractory patient, and eliminating autoreactive B cells that may not have been depleted by other therapies fits into most pathogenic constructs of SLE. Nonetheless, it would be prudent to study anti-CD20 biologics in refractory LN in a well-controlled trial of patients having uniformly defined refractory disease. The role of newly approved LN drugs and drugs that are in development remains to be seen. The possibility that molecular evaluation of the kidneys from refractory patients may provide clues to inflammatory pathways not controlled by conventional treatments is exciting, and suggests that in the future patients with refractory disease may be able to be treated more precisely, thereby avoiding immunosuppressive roulette.

Finally, given the ongoing SARS-COV-2 pandemic, it cannot be overstated that many of the approaches to refractory disease will put these patients at high risk for severe infection, and may prevent adequate protection from vaccination. We suggest that pre-exposure monoclonal antibodies be given to refractory patients, and that all other safety measures including social distancing and masking be maintained.