Introduction

Multiple myeloma (MM) has a low possibility of cure. However, important therapeutic improvements have been made. Scheduled autologous hematopoietic stem cell transplantation (auto-HSCT) after initial cytoreduction has resulted in a complete remission/near-complete remission (CR/nCR) rate of about 20% [1], which translates into a benefit in overall survival (OS). Recently, the advent of targeted therapy such as bortezomib results in important advances in the treatment of MM. Bortezomib, a proteasome inhibitor that inhibits nuclear factor kappa B [2], leads to a CR/nCR rate of 10% in heavily treated, refractory, and relapsing patients with MM [3, 4]. Moreover, upfront use of bortezomib in combination with chemotherapy has resulted in a CR rate about of 20% (17–43%) even in the absence of auto-HSCT [5, 6]. This is an important aspect in the management of MM as MM is a disease of the elderly, with the majority of the patients ineligible for auto-HSCT that leads to a high CR rate [1]. Furthermore, bortezomib is non-myelotoxic and hence may be used in combination with conventional chemotherapy. Indeed, synergism of tumor cytotoxicity has been demonstrated when chemotherapy is combined with bortezomib, thus restoring chemosensitivity to previously chemoresistant tumor cells [7, 8]. Based on these encouraging data, the upfront use of bortezomib in combination with conventional chemotherapy has been tested in younger patients before HSCT and in elderly patients who are not HSCT candidates. Expectedly, the upfront use of combined bortezomib and chemotherapy has resulted in a CR rate of about 20–30% in elderly patients [9] and a more encouraging CR rate of about 50% in younger patients when followed by auto-HSCT [10, 11]. The improved CR rates have translated into superior progression-free survival and OS [12].

Both bortezomib and HSCT have emerged as important components in the management of MM, although their roles and timing relative to each other have not been determined. Bortezomib is an expensive medication with significant side effects. To optimally position bortezomib and HSCT, we reasoned that patients reaching a satisfactory response with conventional treatment might not need additional cytoreduction with bortezomib pre-HSCT. However, for patients with suboptimal response to conventional treatment, additional cytoreduction with bortezomib pre-HSCT might improve the outcome. Therefore, a staged approach was adopted in which newly diagnosed myeloma patients who achieved a ≥75% reduction in paraprotein level after standard vincristine, adriamycin, and dexamethasone (VAD) would proceed to auto-HSCT, whereas those with a <75% paraprotein reduction would receive bortezomib, thalidomide, and dexamethasone (VTD) for additional cytoreduction prior to auto-HSCT.

Materials ad methods

Patients, treatment algorithm, and protocol

The study started in October 2005. The median follow-up time was 17 months (range, 7–39 months). Inclusion criteria were newly diagnosed MM with symptoms, age <65 years, measurable disease, serum creatinine <3 times the upper reference value (normal, 82–126 μmol/L), and adequate liver function [15]. All patients received initial cytoreduction with three cycles of VAD. Those achieving ≥75% reduction in paraprotein proceeded to HSC mobilization with cyclophosphamide (4 g/m2 intravenously) and granulocyte colony stimulating factor (300 μg/day subcutaneously unit leukocyte recovery). Patients with <75% reduction in paraprotein received VTD (bortezomib, 1.3 mg m−2 day−1 intravenously on days 1, 4, 8, and 11; thalidomide, 200 mg/day; and dexamethasone, 40 mg/d orally from days 1–4 and days 8–11). Thalidomide and bortezomib dosage was reduced to 75% for World Health Organization grade III toxicity and omitted for grade IV toxicity. Fourteen of the 25 patients proceeded to VTD salvage therapy because of <75% reduction in paraprotein level. Of these 14 patients, all received four cycles of VTD, except one who had only three cycles of VTD due to grade IV neurotoxicity. Auto-HSCT conditioning regimen comprised intravenous melphalan at 200 mg/m2. Four patients did not receive auto-HSCT. Two patients opted for allogeneic HSCT with HLA-identical siblings, and they were censored after completion of VTD. The other two patients had primary refractory disease or rapidly progressive disease during VTD therapy. All patients received thalidomide (100–200 mg/day) as maintenance therapy regardless of whether VTD had been used. Prophylactic cotrimoxazole for Pneumocystis jirovecii was routinely prescribed. The protocol was approved by the institution review board in accordance with the Declaration of Helsinski, and informed consent was obtained from all participating patients. The treatment algorithm was shown in Fig. 1. Toxicity was reported after initial VAD and after VTD according to the Common Terminology Criteria for adverse events v3.0.

Fig. 1
figure 1

Algorithm of patient treatment

Full size image

Staging and laboratory investigations

MM workup included bone marrow examination, skeletal survey, serum β2-microglobulin (β2M) level, serum protein electrophoresis (SPE), urine protein electrophoresis, serum or urine immunofixation, paraprotein level assay, and serum free light chain (FLC) assay (Freelite, The Binding Site, Birmingham, UK) [16]. All sera were assessed with SPE and FLC immunoassays. Urine was assessed for monoclonal FLC by immunofixation.

Response criteria

All patients were analyzed on an intention-to-treat basis. Bone marrow plasmacytosis and paraprotein levels were assessed prior to treatment, after VAD, after VTD, and 3 and 6 months after auto-HSCT. Responses were defined according to standard criteria [17]. CR was defined as complete resolution of disease with absent paraprotein, as evidenced by a negative SPE and immunofixation, and <5% plasma cells in the bone marrow. nCR was defined as a negative SPE but positive immunofixation. Partial response (PR) was sub-classified into very good partial remission (VGPR, paraprotein reduction ≥90%), PR75 (paraprotein ≥75% but <90% reduction), and PR50 (paraprotein ≥50% but <75% reduction). Minor response (MR) was defined as paraprotein reduction of ≥25% but <50%. No response (NR) was defined as paraprotein reduction of <25%. Progression was defined as ≥25% paraprotein increase in two consecutive tests 4 weeks apart. Relapse was defined as reappearance of the paraprotein on immunofixation in CR patients, positive SPE in the nCR patients, and/or appearance of new bone lesions. For patients with light-chain MM (LCMM), CR was defined as normalization of the level and ratio of serum FLC and negative serum and urine immunofixation. Oligoclonal reconstitution was defined as the appearance of a new paraprotein not present at diagnosis, which persisted for ≥4 weeks. Six of the patients developed a paraprotein different from that at diagnosis during CR or plateau phase and, hence, a “clonal change”. However, the origin of the new clone was not clear and might imply development of a second malignancy entirely unrelated to the original disease. Therefore, in the case of oligoclonal reconstitution, the absence of the original paraprotein by serum/urine immunofixation still qualified for CR.

Statistical analysis

The primary endpoint was response rate. Secondary endpoints were survivals. OS was defined as time from commencement of induction therapy to death or last follow-up. Event-free survival (EFS) was defined as time from commencement of induction therapy to the date of progression, relapse, or death. Survival curves were plotted by Kaplan–Meier method.

Results

Patients

The demographics of the patients were shown in Table 1. There was an incremental upgrade of response after each stage of treatment (Table 2). After VAD, 14 (56%) proceeded to receive VTD. The CR rate was 4% after VAD, 8% after VTD, and 48% in an intention-to-treat analysis after HSCT, or 57% for patients who actually completed HSCT. All patients undergoing auto-HSCT had at least PR75 pre-HSCT. The projected 3-year OS was 75.1% (Fig. 2). Seven patients progressed, all having DS stage III disease with ISS stage II in four and stage III in three cases. The 3-year EFS was 48.3% (median, 26 months; Fig. 2). Four patients died (all with DS stage IIIA disease), one of primary refractory disease, two of relapses after prior CR, and one of progression from PR.

Table 1 Patient demographics (N = 25)
Full size table
Table 2 Cumulative response after each stage of treatment
Full size table
Fig. 2
figure 2

Overall survival and event-free survival of 25 patients

Full size image

Side effects (Table 3)

After VAD, 12 (48%) patients developed grade 2 sensory neurotoxicity, two (8%) tremor (one grade 1, one grade 2), three (12%) with GI side effects (two grade 1 constipation, one epigastric pain), and seven (28%) infective/febrile episodes. After VTD, 11 (78.6%) had sensory neuropathy (grade 3, N = 4; grade 4, N = 1). Of these 11 VTD patients with sensory neuropathy, seven already had grade 2 sensory neuropathy after VAD, of which three had static grade 2 sensory neuropathy and four had increased severity of neuropathy (neuropathy increased from grade 2 to 3, N = 3, and grade 2 to 4, N = 1). Another four patients with no neuropathy after VAD developed neuropathy after VTD (grade 2, N = 3; grade 3, N = 1).

Table 3 Side effects after VAD and VTD
Full size table

Oligoclonal reconstitution

Six patients developed oligoclonal reconstitution, five after CR (as defined by disappearance of the initial paraprotein; Table 4). Three patterns were observed. Cases 1, 2, and 6 with initially LCMM developed a complete immunoglobulin paraprotein. Case 5 developed an additional paraprotein with the same immunoglobulin subtype. Finally, cases 3 and 4 developed an entirely different paraprotein. For patients with a measurable new paraprotein, the level appeared to be stable over a period of 1–11 months (median, 5 months; Fig. 3).

Table 4 Oligoclonal reconstitution in six myeloma patients
Full size table
Fig. 3
figure 3

Oligoclonal reconstitution in MM patients. Upper left showed clonal change of free kappa at diagnosis to IgGκ during complete remission (CR). Upper right showed clonal change from IgAκ at diagnosis to IgGκ during CR. Lower left showed clonal change from IgDλ (arrows) plus free λ (arrowhead) at diagnosis to IgGκ during CR. Lower right showed clonal change from single IgGκ at diagnosis to double IgGκ during CR

Full size image

CNS relapse

Patient P4 (Table 5) with stage IIIA IgA MM achieved PR50 after VAD and CR with VTD. Acute spinal cord compression developed a year after HSCT, with IgA increasing to 2,020 mg/dL (normal <386 mg/dL). Magnetic resonance imaging showed an extensive thoraccolumbar extradural mass with cord compression (Fig. 4a). He achieved a partial response after bortezomib-CMP (cyclophosphamide, melphalan, and prednisolone) [8], but died of disease progression 2 months afterwards. Patient 3 with DS IIIA MM achieved PR75 after VAD, which was maintained with thalidomide after HSCT. Sudden diplopia due to right abducens nerve palsy and skin nodules developed 7 months after HSCT. Computerized axial tomography did not reveal any abnormality. The cerebrospinal fluid showed elevated protein 1.36 g/L (normal < 0.6 g/L) and a cell count of 52 × 106/L with 96% plasma cells (Fig. 4b). The serum paraprotein level resurged to 27 g/L. Bone marrow examination showed 28% pleomorphic plasma cells. He refused further treatment and died 4 weeks afterwards.

Table 5 Characteristics and outcome of the seven patients with disease progression
Full size table
Fig. 4
figure 4

a Sagittal fat-saturated SE T1W scan obtained after administration of intravenous gadolinium contrast agent showing an enhancing epidural mass (arrows) in the posterior spinal canal compressing the thoracic cord. b Cerebrospinal fluid showed myeloma cells

Full size image

Discussion

In this study, the response to standard VAD was used to dichotomized patients to a chemosensitive group (≥75% paraprotein reduction after VAD) who proceeded directly to auto-HSCT and a relative chemoresistant group (<75% paraprotein reduction) who received additional cytoreduction with VTD before auto-HSCT. We surmised that this staged approach might maximize the reduction of neoplastic cells prior to HSC mobilization and restrict the use of the expensive bortezomib to chemoresistant patients.

There are several interesting observations. We achieved a cumulative CR rate of >50% after auto-HSCT. Recent studies showed that first-line use of bortezomib-based regimens gave CR rate of up to 30% even without HSCT [9]. Furthermore, total therapy using bortezomib-containing regimens for initial cytoreduction prior to auto-HSCT resulted in a CR rate of about 50% (43% to 56%) [10–12]. Therefore, the results in this study provide an alternative method of achieving an appreciable CR rate with a reduced use of bortezomib. In less affluent countries or health care systems, our approach offers a means of alleviating the financial burden of bortezomib without compromising the overall treatment results of patients with MM.

Despite the relatively high CR rate after auto-HSCT in our cohort, it is noteworthy that the CR rate only changed from 4% to 8% when VTD was used in patients not responding optimally to VAD. These results contrasted that of the regimen (PAD, bortezomib, doxorubicin, dexamethasone) in untreated MM patients where the CR rate was around 25% after PAD alone [10, 11]. While these results could not be directly compared as different regimens were used on dissimilar patient populations, this disparity might be partially explained by the fact that only relatively chemoresistant patients received VTD in our study.

On the other hand, using VAD upfront resulted in very frequent (48%) sensory neuropathy prior to VTD. Moreover, eight had new onset or increased severity of neuropathy after VTD. Therefore, it would have been more preferable to use AD instead of VAD to avoid excessive neuropathy without compromising the chemotherapy response prior to VTD.

On follow-up, a significant number of patients developed a new paraprotein distinct from the one at diagnosis. Indeed, development of isotypic change and oligoclonal bands has been reported in about 10% of MM patients undergoing HSCT [18]. The long-term outcome of these new clones remains to be determined. If these new clones represented expansion of minor subclones consequent to effective elimination of major clones, this phenomenon of oligoclonal reconstitution might become more common as more effective treatment becomes available. Alternatively, appearance of the new clone might represent clonal evolution from the original clone. Case 5 who developed two Gκ bands evolving from one Gκ at diagnosis might be explained by this mechanism. Whatever the pathogenesis, oligoclonal reconstitution affects the management of patients. Firstly, the diagnosis of CR may be difficult. In our case, we define CR as the absence of the original clone, and hence, the emergence of another clone does not alter the diagnosis of CR. Secondly, the long-term outcome of these new clones in our patients is yet to be defined, which might be transient [18].

Two of our patients developed central nervous system (CNS) myeloma at progression in addition to cutaneous deposition and plasma cell leukemia. This is a rare complication [19, 20], and the factors predisposing to CNS dissemination remain to be elucidated. However, it would have been ideal if clonal analysis of the CSF plasma cells had been performed.

Finally, deep vein thrombosis (DVT) did not occur in any of our patients despite the omission of prophylaxis. In Western patients, when thalidomide was combined with dexamethasone or chemotherapy, the risk of DVT might be increased to 10–27% [21]. Our observation concurred with another study showing a low incidence of DVT in a cohort of 85 Chinese MM patients receiving thalidomide [22]. Moreover, the incidence of DVT in Chinese has been shown to be much less frequent than Caucasian patients to the extent that prophylaxis is not a standard practice even in patients undergoing orthopedic surgery [23, 24]. This disparity might at least be partially explained by the absence of factor V Leiden and prothrombin 20210 mutations, both hereditary thrombophilic tendencies prevalent in Caucasians, in the Chinese population [25, 26]. Therefore, we believe that Chinese patients do not require prophylactic aspirin while receiving thalidomide.

In conclusion, our VAD-VTD/auto-HSCT algorithm resulted in a high response rate and promising survivals. This staged approach has significant financial implications in the treatment of MM.