The kidney and monoclonal free light chains

From Wikilite

13.1. Introduction

Renal failure is a major cause of morbidity and mortality in patients with MM. At initial presentation, up to 50% of patients have renal impairment (serum creatinine >1.5mg/dL or >130µmol/L); 12 to 20% have acute renal failure (ARF) and 10% become dialysis dependent. They represent 2% of the dialysis population and there are approximately 5,000 new patients, worldwide, each year. Furthermore, there is a 50% mortality within 6 months of diagnosis ^[1][2].

While reversible factors such as dehydration, hypercalcaemia and medication are frequently involved, monoclonal FLCs are the most potent cause of irreversible renal failure. Large amounts of sFLCs readily pass through the glomerular fenestrations and overwhelm the absorptive capacity of the proximal tubules. On entering the distal tubules, they co-precipitate with Tamm-Horsfall protein to form waxy casts that both block the flow of urine and cause interstitial inflammation ^[3]. Furthermore, high concentrations of FLCs are directly toxic to tubular cells ^[4][5].

Studies have analyzed renal recovery rates after FLC removal by plasma exchange. This is a logical approach, but results have been disappointing. Although an early report was optimistic ^[6], the largest and most recent controlled trial (97 patients) showed no clinical benefit ^[7]. A subsequent editorial in the Journal of the American Society of Nephrology (JASN) listed the shortcomings of this study, including the failure to monitor either serum or urine FLC concentrations ^[8]. It was noted, “This resembles anti-hypertensive treatment without measuring blood pressure.” Clearly, the efficiency of plasma exchange for FLC removal could not be judged.

Because FLCs are relatively small protein molecules (κ~25 kDa: dimeric λ~50 kDa) they are present in similar concentrations in serum, extravascular compartments and tissue oedema fluid ^[9]. Thus, the intravascular compartment may contain only 15 to 20% of the total amount. A series of 3.5 litre plasma exchanges that removed 65% of intravascular FLCs on each occasion would have little overall impact, particularly if production were not reduced at the same time by chemotherapy. An alternative approach is to remove FLCs by haemodialysis. Although this is not possible with routine dialyzers because of their small pore sizes (12-15kDa), a new generation of “high cut-off” dialyzers, allows FLC removal ^[2]. By using extended dialysis, large amounts of FLCs can be removed without the attendant clotting and deproteination problems that may limit the extended use of plasma exchange.

This chapter discusses the normal renal handling of FLCs, their role in renal failure in MM, clinical case studies and current management strategies such as plasma exchange. A mathematical model of FLC removal is presented and plasma exchange is compared with the utility of haemodialysis. Finally, clinical evidence for the beneficial use of “FLC removal haemodialysis” is presented.

13.2. Normal free light chain clearance and metabolism

Figure 13.1 Renal injury caused by FLCs. (This figure was published in Acute renal failure: Myeloma kidney. Winearls CG. In: Johnson RJ, Feehally J, eds. Comprehensive clinical nephrology, Mosby: Page 238, figure 17.5 © Elsevier (2003)).

In normal individuals, sFLCs are rapidly cleared by the kidneys depending upon their molecular size (Figure 13.1 and Chapter 3). Monomeric FLCs, characteristically κ, are cleared in 2-4 hours at 40% of the glomerular filtration rate. Dimeric FLCs, typically λ, are cleared in 3-6 hours at 20% of the glomerular filtration rate, while larger polymers are cleared more slowly. Removal is prolonged to 2-3 days in MM patients who are in complete renal failure when FLCs are removed by the liver and other tissues. In contrast, IgG has a normal serum half-life of 21 days that is not affected by renal impairment.

After filtration by the glomeruli, FLCs enter the proximal tubules and bind to brushborder membranes via low-affinity, high-capacity receptors called cubulins and megalins ^[10]. Binding provokes internalisation of the FLCs, subsequent proteolysis into smaller peptides and finally their excretion into the urine flow. The concentration of FLCs leaving the proximal tubules, therefore, depends upon the amount in the glomerular filtrate, competition for binding uptake from other proteins and the absorptive capacity of the tubular cells. A reduction in GFR, due to loss of nephrons, increases sFLC concentrations so that more is filtered by the remaining functioning nephrons. Subsequently, and with increasing renal failure, hyperfiltering glomeruli leak albumin and other proteins which compete with FLCs for absorption thereby causing more to enter the distal tubules.

FLCs entering the distal tubule can bind to uromucoid (Tamm-Horsfall protein). This is the predominant protein in normal urine and is thought to be important in preventing ascending urinary infections. It is a glycoprotein (85kDa) that aggregates into high molecular weight polymers of 20-30 units. Interestingly, it contains a short peptide motif that has a high affinity for FLCs ^[11].

13.3. Nephrotoxicity of monoclonal free light chains

Figure 13.2 A Waxy cast from the urine of a patient with multiple myeloma. (This figure was published in Investigation of renal disease: Urinalysis. Fogazzi GB. In: Johnson RJ, Feehally J, eds. Comprehensive clinical nephrology, Mosby: Page 41, figure 4.3B © Elsevier (2003)). B Classic casts in the distal tubules of a patient with light chain multiple myeloma.

Figure 13.3 A Serum λ FLC concentrations in a patient with IgAλ MM during development of acute renal failure. B Serum FLC concentrations in a patient with LCMM during the development of acute renal failure and subsequent response to VAD (arrow). (Courtesy of S Abdalla).

Figure 13.4 κFLCs causing renal impairment in a patient who appeared to have stable disease from IgG measurements.Satisfactory improvement in renal function followed chemo-therapy. VED: vincristine, epirubicin and dexamethasone. (Courtesy of MR Nowrousian).

The main pathology in myeloma kidney is cast nephropathy. This is caused by precipitation of FLCs with uromucoid as waxy casts and is characteristically found in acute renal failure associated with MM (Figures 13.2A and 13.2B) ^[12]. The casts obstruct tubular fluid flow, leading to disruption of the basement membrane and interstitial damage. Rising concentrations of sFLCs are filtered by the remaining functioning nephrons which become blocked, leading to a vicious cycle of further increases in sFLC concentrations and progressive renal damage. This may explain why some MM patients, without apparent pre-existing renal impairment, suddenly develop catastrophic and irreversible renal failure. The process is aggravated by other factors such as dehydration, diuretics, hypercalcaemia, infections and nephrotoxic drugs. Monoclonal FLCs cause renal impairment by several mechanisms, any of which may contribute to both acute myeloma kidney and chronic renal failure.

1. Activation of inflammatory mediators in the proximal tubule epithelium.
   
2. Proximal tubule necrosis.
   
3. Fanconi syndrome (renal tubule acidosis) with FLC crystal deposition.
   
4. Cast nephropathy.
   
5. AL amyloidosis (Chapter 15).
   
6. Light chain deposition disease (Chapter 17).
   

In MM, monoclonal sFLC concentrations are abnormal in >95% (Chapter 11) and can have a wide range of concentrations. Importantly, their toxicity also varies considerably. This was elegantly shown by Sanders and Brooker ^[4][5] using isolated rat nephrons and is, in part, related to variable affinity for Tamm-Horsfall protein. However, in spite of much effort to show otherwise, particular molecular charge and κ or λ type are not now considered relevant to FLC toxicity. Furthermore, highly polymerised FLCs (a frequent finding in MM - (Chapter 4)) are probably not nephrotoxic because they cannot pass through the glomerular fenestrations. This may partly account for the lack of renal damage in some patients who have very high sFLC concentrations.

The amount of sFLCs necessary to cause renal impairment was studied by Nowrousian et al ^[13]. They showed that the median serum concentrations associated with overflow proteinuria (and hence potential for tubular damage) was 113mg/L for κ and 278mg/L for λ. These are approximately 5-10 fold above the normal serum concentrations and presumably relate to the maximum tubular reabsorption capacity of the proximal tubules. Since the normal daily production is ~500mg, increases to ~5g/day are likely to be nephrotoxic in most patients.

As renal impairment develops, progressive increases occur in both the monoclonal sFLCs and the polyclonal non-tumour sFLCs (Chapter 20). Concentrations of monoclonal sFLCs below 300mg/L are rarely associated with renal impairment as judged by the associated normal levels of the non-tumour FLCs (Figure 8.2). These concentrations are somewhat higher than those observed by Nowrousian et al. ^[13], in patients with renal impairment, but are in the same general range.

There have been several urine studies that have related urine FLC excretion rates to renal impairment. Typically, the associated renal impairment rises with increasing urine FLCs. One study showed that 7%, 17% and 39% of patients had renal impairment with excretions rates of <0.005g/day, 0.005-2.0g/day, and >2g/day, respectively ^[14]. However, FLC excretion is an indicator of renal damage in addition to its cause.

The causal relationship between serum FLC concentrations and renal impairment is illustrated in 3 patients (Figures 13.3 to 13.4) ^[15]. In each case, urine FLC concentrations were low and not indicative of the underlying renal deterioration.

The patient shown in Figure 13.3A had developed MM expressing IgAλ and λ FLCs, 3 years earlier. She had a good initial response to VAD but was frail. Over a 7-month period, her serum λ FLC levels increased from 1,500 to 2,800 mg/L, then 4,500 mg/L and finally 15,000mg/L in month 9, while serum creatinine had increased slightly to 120 µmol/L. Her medication was not changed from thalidomide, cyclophosphamide and pulsed dexamethasone on account of her frailty. Unfortunately, the sharp rise in serum FLCs led to acute renal failure which was apparent one month later (as shown by the high creatinine concentration). She failed to respond to therapy and died shortly after.

Figure 13.3B illustrates a patient who developed acute renal failure that responded to chemotherapy and haemodialysis. As in Figure 13.3A, the serum FLC concentrations were a sensitive indicator of impending renal failure and concentrations appeared to reach a threshold before acute renal failure developed. Although renal function was deteriorating at the 4th month, the patient felt well and declined treatment, only to present in renal failure one month later. The potential seriousness of the high FLC concentrations had not been fully appreciated. She was given VAD and put on haemodialysis for 6 weeks. This produced a good response of the MM but only partial renal recovery. It is of note that sFLC concentrations were, consistently, more sensitive than serum creatinine as a marker of falling GFR.

Figure 13.4 shows sFLCs, immunoglobulins and creatinine results in another patient over a 30-week period. Total protein and IgG tests indicated stable disease during the first 18 weeks of monitoring. However, the serum creatinine increased from 1.2 to 3.0mg/dL (130 to 240 µmol/L). sFLC measurements showed unexpectedly high κ concentrations at 500 mg/L with a κ/λ ratio of 4.04, in spite of no FLCs in the urine. It was considered that the high FLC concentrations might be causing renal damage so chemotherapy was commenced. Over the ensuing 2 months the serum κ and serum creatinine almost returned to normal concentrations.

13.4. Diagnosis of myeloma kidney using sFLC analysis

Figure 13.5 sFLCs in 142 patients presenting with dialysis-dependent, acute renal failure. 41 patients with κ (triangles) and λ (squares) multiple myeloma are distinct from blood donors (red crosses: P<0.001) and from patients with ARF from other causes (diamonds: P<0.001).

Figure 13.6 Calculated reduction of κ sFLCs in a patient with LCMM by 100% (immediate) tumour killing (1) and 10% per day (2). The effect of 6 x 3.5 litre plasma exchanges over 12 days is indicated (3) based on 10% kill per day (2) plus the addition of haemodialyis for 4 hours, x3 per week (4) and 8 hours/day (5).

The diagnosis of renal failure from MM has been based upon testing for FLCs in the urine together with SPE and serum IFE. Positive results are followed by a renal biopsy. Since sFLC tests are more sensitive than urine tests, it is logical to commence with sFLC analysis alone. SPE and sIFE are irrelevant since they test for intact monoclonal immunoglobulins, which do not cause myeloma kidney.

To confirm the diagnostic accuracy of sFLC analysis we reviewed an unselected cohort of 142 patients who had presented with dialysis-dependent acute renal failure of unknown cause. Of the 44 patients with monoclonal gammopathies, 41 had MM with abnormal sFLC ratios and 1 had MGUS (Figure 13.5). Receiver operating characteristic (ROC) curve analysis showed that a slightly modified sFLC ratio range increased the specificity from 93% to 98% with no loss of sensitivity (Chapter 20). By comparison, urine testing for monoclonal FLCs was negative in one of the 24 MM patients that were assessed ^[16] and 17 had inadequate urine samples.

These results indicate that sFLC κ/λ ratios are a specific, yet simple method for identifying monoclonal FLC production in patients with MM and acute renal failure. Rapid diagnosis in these patients will allow early initiation of renal biopsies for confirmation of myeloma kidney followed by “high cut-off haemodialysis” and the appropriate chemotherapy (Figure 13.6).

13.5. Removal of free light chains by plasma exchange

Clearly, the pre-renal load of sFLCs is an important factor in renal toxicity. It seems logical that renal recovery might occur if sFLC concentrations were lowered rapidly. Hence, it has been suggested that plasma exchange might improve the current poor outcome of these patients.

Several early studies indicated renal function improvement using plasma exchange. For example, in 1988, Zucchelli et al. ^[6], compared MM patients on peritoneal dialysis (control group) with plasma exchange (some patients also on haemodialysis). Only 2 of 14 in the control group had improved renal function compared with 13 of 15 in the plasma exchange arm, and survival was better (P<0.01). However, in this setting compared with haemodialysis, peritoneal dialysis is associated with a worse overall survival making the comparison not ideal.

This early success was not repeated in subsequent controlled trials. WJ Johnson et al. ^[12], in 1990, compared 10 patients on forced diuresis with 11 who had forced diuresis and plasma exchange and found no difference in outcome. Most recently, a large series was reported by WF Clark et al. ^[7], (2005). Patients were treated with chemotherapy, haemodialysis or a combination of the two, and were randomly allocated to receive additional plasma exchange. Again, there was no statistically significant benefit for plasma exchange.

13.6. Model of FLC removal by plasma exchange and haemodialysis

Figure 13.7 Failure of 16 plasmapheresis sessions (arrows) to lower sFLCs (circles) but decline in response to Bortezomib (B), Cyclophosphamide (C): Dexamethasone (Dex) and Thalidomide (Thal). (Reproduced with permission from Transfusion and Cserti ^[17]).

Figure 13.8 Photomicrograph of pores in a Gambro high “cut-off” (HCO™) membrane and a high-flux membrane (Polyflux™). (Courtesy of H Goehl, Gambro Dialasoren GmbH).

In order to understand the effectiveness of plasma exchange we developed a compartmental mathematical model that was applicable to patients being treated for MM and renal failure ^[2]. The following parameters were considered:- sFLC concentrations at clinical presentation; monomeric κ and dimeric λ clearance rates with and without renal failure; partition of sFLC between vascular and extravascular compartments (including oedema fluid); flow of FLCs between compartments; half-life of sFLC in renal failure, sFLC production rates, and tumour killing rates with chemotherapy. Data from patients with MM were fitted to the model to analyse rates of sFLC removal. Simulations were conducted to compare 6 and 10 plasma exchange treatments (of 3.5L each over 12 days) with 5 different haemodialysis protocols. Chemotherapeutic tumour killing rates that were considered included 100% on the first day; 10%, 5%, and 2% per day and no killing.

In renal failure the half-life of sFLCs is approximately 3 days. Assuming a starting concentration of 10,000mg/L, it would take approximately 20 days for the FLCs to be metabolised assuming complete tumour killing with the first chemotherapy dose (Figure 13.6: 1). More realistically, 10% of the tumour might be destroyed per day by aggressive chemotherapy resulting in a slower FLC reduction (Figure 13.6: 2). Addition of plasma exchange procedures increased removal rates by approximately 25% but concentrations were not reduced below toxic levels (500mg/L) at 4 weeks. The rapid reductions in serum FLC concentrations during the procedure and their subsequent reentry from the extra-vascular to the intra-vascular compartment can be seen.

Results from the model calculations suggest that plasma exchange is not effective unless used intensively. One report on 3 patients undergoing plasma exchange suggested no benefit (Figure 13.7) ^[17], while another report, on one patient, considered it was helpful ^[18]. Lack of success can be explained by on-going high production rates and re-entry of FLCs from extravascular compartments (including oedema fluid).

It should be noted that peritoneal dialysis does not remove FLCs efficiently (Chapter 20). Presumably, this is because the volume of exchanged fluid is much lower than in haemodialysis.

13.7. Removal of free light chains by haemodialysis

Figure 13.9 Calculated clearance of serum kappa FLCs with 10% tumour killing per day (2) and the addition of haemodialyis for 8 hours/day (5), 12 hours per day (6), 8 hours on alternate days but no tumour killing (7) and no tumour killing with no haemodialysis (8). (See Table 13.1 for calculation details and Figure 13.6).

Figure 13.10 Concentrations of λ FLCs in serum and dialysate fluid of a patient with acute renal failure due to LCMM undergoing dialysis with the Gambro “high cut-off” dialyser. The dialyser was renewed (arrows) as FLC leakage slowed.

As an alternative to plasma exchange, sFLCs might be removed more effectively by haemodialysis, provided the pore sizes of the membranes are large enough. Conventional dialysers have a molecular weight cut-off around 10-15kDa so the filtration efficiency for FLCs is very low. However, some of the new “protein-leaking” dialysers have much larger pores, particularly the Gambro “high cut-off dialyser” (Figure 13.8) ^[2]. Furthermore, haemo-diafiltration is more effective at removing small protein molecules than haemodialysis. Indeed, there are sporadic reports of patients with AL amyloidosis and end-stage renal failure who appear to have improved survival when haemo-diafiltration is instigated. But, there is currently inadequate clinical data to provide a clear conclusion in amyloidosis patients.

Model calculations (Figures 13.6 and 13.9 and Table 13.1) indicate that the prolonged use of “high cut-off” dialysers could reduce κ sFLC concentrations to less than 0.5g/L in 2-3 days with ~95% of the sFLC removed. Dimeric λ molecules are considerably larger so they are removed more slowly. Based upon these modelling results, we assessed the utility of various dialysers for serum FLC removal in patients with MM. The Gambro high cut-off membrane with pores that are large enough to filter FLCs was the most efficient by a considerable margin. Figure 13.10 shows the rapid removal of λ FLCs in a patient with MM using the Gambro HCO™ dialysis membrane over a 6 hour period. Also shown are the large amounts of FLCs in the dialysate fluid. A total of nearly 40 grams of FLCs were removed in a single dialysis session.

Table 13.1. Model calculations of the efficiency of therapeutic removal of sFLCs. Numbers are the additional percentage of FLC removed by intervention compared with normal metabolism. Numbers in parentheses are the time in days for serum FLC concentrations to reduce to 5% of the starting concentrations (from 10g/L to 0.5 g/L). For superscript numbers 1 to 8, the simulations are shown in Figures 13.6 and 13.9. aSerum FLC concentrations at day 150 for simulations in which reductions to 0.5 g/L were not achieved. PE: plasma exchange. HD: hemodialysis. NA: not applicable.

13.8. Recovery from renal failure following FLC removal haemodialysis

Figure 13.11 sFLCs in a patient presenting with MM and acute renal failure who responded to haemodialysis with renal recovery. Pre- and post-dialysis concentrations are shown. Numbers indicate hours of dialysis and FLCs removed (in grams) per session. Dialysis was over 22 days leading to renal recovery that has been maintained for over 2 years.

Figure 13.12 sFLC in a patient with relapsing MM and ARF who recovered with chemotherapy and haemodialysis. Pre- and post-dialysis sFLC measurements are shown. The patient received dialysis over 20 days then had renal recovery with GFR rising to 50. At 1 year the patient remained in complete MM remission.

Figure 13.13 sFLCs in a patient presenting with relapsing MM and acute renal failure who remained dialysis-dependent, in spite of removing 1.7 Kg of FLCs (shown per 10 day period).

Figure 13.14 A Kaplan-Meier plot of renal recovery in 17 patients with MM and acute renal failure treated by chemotherapy and sFLC removal haemodialysis. The control group comprised 17 patients with MM treated over the previous 6 years with similar chemotherapy but using conventional haemodialysis 3 times per week. B Kaplan-Meier plot of the 12 patients with acute renal failure who had recovery of MM following chemotherapy (from Figure 13.6 above). The control group comprised 7 patients with MM treated over the previous 6 years who went into remission but who were treated with conventional haemodialyis x3 per week.

Figure 13.15 Kaplan-Meier plot of survival time in 17 patients with MM and acute renal failure treated by FLC removal haemodialysis. Those with renal recovery had longer survival than those who remained on haemodialysis.

Figure 13.16 Renal biopsies of a patient A, at presentation and B, 6 weeks after “high cut-off” haemodialysis showing resolution of FLC casts in the distal tubules. (Courtesy of K Basnayake).

Intensive application of FLC removal haemodialysis has been applied to 17 patients with cast nephropathy by Hutchison et al. ^[19], in Birmingham, UK. Figures 13.11 to 13.13 show examples of the clinical response in 3 patients who presented with acute renal failure. The first of the 3 figures shows good FLC reductions with renal recovery (Figure 13.11. The second failed to respond to the initial chemotherapy but responded to bortezomib (Figure 13.12). Haemodialyis was unable to produce sustained FLC responses without accompanying satisfactory chemotherapy. The third patient failed to achieve renal recovery (Figure 13.13). He developed sepsis and chemotherapy was stopped. In spite of multiple, prolonged periods of haemodialysis and removal of 1.7Kg of FLCs, there was no recovery of renal function. sFLC concentrations oscillated during treatment in much the same manner as shown from model calculations shown in Figure 13.9 (line 7).

It should be emphasised that all patients needed both effective chemotherapy to switch off FLC production together with rapid reduction of the pre-renal load, for renal recovery to occur. Dialysis independence occurred in patients who achieved a >60% sustained reduction in sFLCs. The 25% FLC reduction achieved using standard plasma exchange is presumably inadequate for renal recovery.

The renal recovery rate of 17 treated patients is shown in Figure 13.14A, alongside historical recovery rates at the same hospital. Clearly, the patients receiving intensive FLC removal had better renal recovery rates. All 12 patients had renal recovery when using FLC removal haemodialysis if they also had tumour responses (Figure 13.14B). This was in contrast to the control group, where only 1 patient had renal recovery out of 7 who responded to chemotherapy. This suggests an additional benefit of dialysis over chemotherapy alone.

Figure 13.15 analyses patient survival in the 17 patients treated with FLC removal haemodialysis. The 5 patients in the treated group who died were those that failed to regain renal function. All 12 patients who became independent of renal replacement therapy remained alive at 6 months. GFR continued to improve for several months after dialysis independence was achieved. Furthermore, this was associated with complete disappearance of the casts in the one patient in whom serial renal biopsies were available (Figure 13.16).

Two German groups have obtained similar results using the same dialysis protocols, and one has reported results from the first 3 patients treated ^[20]. Overall, these exceptional results suggest that renal recovery is possible in most patients with MM presenting with acute renal failure from cast nephropathy.

Other groups have suggested that high rates of renal recovery are possible with modern chemotherapy alone, particularly using combinations that include bortezomib. Ludwig et al. ^[21], showed recovery in 40% of patients, but none had renal biopsies and few were dialysis dependent. Even better recovery rates, at 73% of patients, were reported by Kastritis et al. ^[22], but again patient selection was unsatisfactory ^[23]. The control group in the plasma exchange study of WF Clark et al., also had 40% renal recovery and were in a similar clinical state. This suggests that few patients with cast nephropathy that are dialysis dependent will recover with chemotherapy alone. Supporting evidence comes from a retrospective study that showed only 25% recovery in dialysis dependent patients with cast nephroapthy given bortezomib and conventional haemodialysis ^[24].

The clinical challenge will be to identify and quickly treat patients after diagnosis. Traditionally, acute renal failure due to MM has been identified by serum and urine studies for monoclonal FLCs. Our data suggests that all patients with MM can be identified using sFLC tests alone (Figure 13.5) and concentrations are usually greater than 500mg/L ^[16]. Renal biopsy is also essential to specifically identify cast nephropathy. Other causes of acute renal failure in MM do not have huge concentrations of sFLCs and do not require intensive, “high cut-off,” haemodialysis for renal recovery.

Return of high GFR rates is likely to be better with rapid reduction of sFLCs to normal concentrations. We have experience of 2 patients whose FLC removal haemodialysis was not commenced until one month after initial presentation with acute renal failure. Recovery to dialysis independence was slow with barely satisfactory GFR at 6 months, in spite of complete MM remission.

A German and UK, multi-centre, prospective, randomised controlled trial comparing the Gambro HCO™ dialyser with conventional haemodialysis is currently underway. It is planned to recruit 90 MM patients presenting with acute renal failure and cast nephropathy. They will receive bortezomib, dexamethasone and doxorubicin with randomisation to intensive FLC removal haemodialysis or conventional haemodialysis ^[25].

References