Background

Globally, at least 5–10 million people are estimated to be infected with Human T-lymphotropic virus type 1 (HTLV-1), of whom ~ 5% develop adult T-cell leukemia/lymphoma (ATL) [1, 2]. ATL is an aggressive CD4+ T-cell malignancy associated with poor prognosis and, despite advances in chemotherapy and supportive care, the overall survival (OS) has not improved in the 35 years since HTLV-1 was first described [3, 4], with a median OS of 9–13 months for the aggressive subtypes (acute and lymphoma) and 4.1 years for the indolent subtypes (chronic and smoldering) [5,6,7]. Allogeneic bone marrow transplantation has been regarded as the only curative therapy [8]. In Europe and the USA, combination treatment with zidovudine and interferon-α (ZDV/IFN-α) has substantially changed the management of patients with the leukemic subtypes of ATL (acute or unfavorable chronic ATL; unfavorable defined by presence of either elevated LDH, elevated serum urea, reduced serum albumin) [9,10,11]. An international meta-analysis reported that patients with chronic or smoldering ATL who received first-line ZDV/IFN-α (n = 17) had an excellent survival (100% OS beyond 5 years) compared with patients who received first-line chemotherapy (n = 6) with or without maintenance ZDV/IFN-α (5 year OS 42%) [11]. By comparison, poor long-term outcomes are reported from watchful waiting: From a single-center study, of 90 patients with indolent ATL (favorable chronic or smoldering), 78 patients were managed by watchful waiting, with a median survival time of 4.1 years with no plateau in the survival curve. Of the 90 patients, 49.9% progressed to aggressive ATL, with a median time of transformation at 18.8 months. Furthermore, those patients treated early with chemotherapy had an OS inferior to those managed with watchful waiting [7].

Whilst long-term remissions has been reported with ZDV/IFN-α, the optimal duration of this treatment is unknown. Anecdotal cases where therapy is discontinued due to side effects or compliance have been associated with rapid disease relapse, and it has been widely accepted that patients require life-long therapy. We report the case of a patient who was diagnosed with chronic unfavorable ATL and treated with ZDV/IFN-α for 5 years. Treatment was discontinued because of toxicity, and the patient remains in a hematological and molecular remission more than 6 years off all therapy.

Case report

A 33-year-old Afro-Caribbean man was diagnosed with chronic ATL in December 2002 following a routine complete blood count (CBC): White cell count (WCC) 13.6 × 109/L, hemoglobin (Hb) 169 g/L, platelets 125 × 109/L, lymphocytes 6.2 × 109/L, normal calcium and lactate dehydrogenase (LDH). Blood film identified atypical lymphocytes which expressed CD2, CD4, CD5 and CD25, but CD3 dim, and CD7 absent. Additionally, the tumour phenotype was CCR4+, FoxP3+, PDL1 negative and Tax negative (type 2 defective provirus) [12]. CT scan was unremarkable. HTLV-1 serology was positive and the HTLV-1 proviral load (PVL), the proportion of HTLV-1 infected peripheral blood mononuclear cells (PBMCs), was 60.3%.

The patient was asymptomatic and initially managed with a watchful waiting approach until September 2004 when he developed unfavorable features with a rise in the LDH at 515 U/L (NR 50–450), and a rise in PVL to 100% (Fig. 1a). Between September 2004 and November 2005, he was treated with a histone deacetylase inhibitor, sodium valproate, with no significant change in WCC or PVL. In November 2005, he commenced zidovudine 500 mg twice daily and 6 MIU interferon-α subcutaneously once daily. Within 5 weeks the WCC had normalized, zidovudine was reduced to 250 mg twice daily and interferon-α switched to a pegylated version (Pegasys 135–180 mcg once weekly). HTLV-1 PVL fell from 106 to 36.5% following 4 months’ treatment. By 2010, HTLV-1 PVL had fallen further, to below 10%. LDH remained mildly elevated (< 2 upper limit normal, ULN) throughout the ZDV/IFN-α treatment. In February 2011, following 62 months of treatment, he developed CTCAE grade 4 transaminitis. Abdominal ultrasound was unremarkable and liver biopsy did not identify evidence of ATL or other cause. ZDV/IFN-α was discontinued and he was treated with a short course of oral prednisolone, following which his liver function tests normalized. At the time of ZDV/IFN-α discontinuation, his CBC was normal and HTLV-1 PVL was 6% and a watchful waiting approach was resumed. In March 2017, 72 months off all treatment, he remains in clinical remission with a normal blood count, LDH and stable HTLV-1 PVL at 10%. Detailed longitudinal clonality analysis was performed to quantify the abundance of all HTLV-1-infected T-cell clones and to monitor the abundance of the presumed malignant clone while on and off therapy.

Fig. 1
figure 1

a White cell count, lymphocyte count and PVL. Green bar indicates the period where sodium valproate was administered, and red line indicates the period where ZDV/IFN was administered. b Absolute abundance of the dominant UIS per 100 PBMCs determined by LM-PCR + HTS (red) and UIS-specific qPCR (blue). The limit of detection of the qPCR assay is indicated by the dotted line. N.D. not detectable by LMPCR. c Relative abundance of the dominant UIS as determined by qPCR (blue line) per 100 proviral genomes. d Relative abundance of dominant UIS (red) within total proviral UIS, at the same time points as b (red). October 2003, untreated; May 2005–October 2005, sodium valproate; November 2005, at commencement of ZDV/IFN treatment; March 2015, off treatment 5 years (10 years post diagnosis)

Full size image

Methods

Diagnosis and monitoring of HTLV-1 infection

Diagnostic tests including PVL monitoring [13] were undertaken in accredited diagnostic laboratories (Imperial College Healthcare NHS Trust and Imperial College London). Written informed consent was obtained from the patient.

Clonal abundance of HTLV-1 Integration sites by high-throughput sequencing (HTS)

As previously reported, a customized protocol of linker-mediated PCR (LM-PCR) followed by high-throughput sequencing (HTS) was used to map and quantify unique HTLV-1 proviral integration sites (UIS) [14].

Absolute abundance of the dominant proviral integration site by qPCR

Primers were designed to amplify the junction between the 3′LTR of the provirus (5′-CAGCGACAGCCCATTCTATA-3′) and the human genome downstream of the dominant proviral integration site (5′-CAGGGGTTCGAACATGAGTT-3′). DNA amplified (FastSybr Sybr green reaction mix, ThermoFisher) on a Viia7 real-time PCR system as per manufacturer’s protocol, normalized to the copy-number of GAPDH or tax [15, 16].

Results

Clonality analysis and residual disease estimation by high throughput proviral integration site mapping

Each clone of HTLV-1-infected cells is defined by a unique genomic integration site of the provirus which can be precisely mapped and quantified in absolute abundance (per 100 PBMCs) and relative to all other HTLV-1-infected cells.

At presentation, HTS detected a dominant proviral UIS at X_147227167 (Chromosome X, position 147227167, human genome build 18) which had an absolute abundance of 21.1 copies/100 PBMC (red dots, Fig. 1b). This increased to 59.6/100 PBMC before commencing sodium valproate and further increased to 84.1/100 PBMC before commencing ZDV/IFN-α. The absolute abundance of the presumed malignant clone declined to 14.7/100 PBMC after 27 months’ treatment; to 0.002/100 PBMC at 54 months’ treatment; to 0.004/100 PBMC at 59 months; and at 5 years off all treatment it was not detected by HTS.

At diagnosis, the UIS X_147227167 constituted 88% of proviral genomes (relative abundance, Fig. 1c). Before starting ZDV/IFN-α treatment, the relative abundance of UIS X_147227167 increased to 99%. When re-analyzed in May 2010, following 4.5 years of treatment, the clone frequency distribution of HTLV-1 infected cells had changed dramatically; the PVL was now composed of many low-abundance clones, of which UIS X_147227167 contributed 0.3% of proviral genomes (Fig. 1d).

Abundance of the dominant clone quantified by integration site-specific qPCR

We corroborated our observations using a clone-specific qPCR which amplified the unique virus/host junction of the presumed malignant clone. By UIS-specific qPCR, the absolute abundance of UIS X_147227167 was 138 copies per 100 PBMC before starting ZDV/IFN-α and remained at this high level during the first 4.5 years on therapy. After this point, the absolute abundance fell progressively to 0.02 copies/100 PBMC, a > 1000-fold reduction. Subsequently, during more than 5 years off treatment, the UIS remained detectable by qPCR, at a low but stable abundance (Fig. 1b). These findings corroborate the data from the high-throughput sequencing protocol (Fig. 1c).

Discussion

We report the clinical case of a patient with chronic unfavorable ATL who responded to ZDV/IFN-α but had to discontinue this treatment after 5 years continuous therapy due to liver toxicity. He has now been off all therapy for 72 months, under active monitoring, and remains clinically well and continues in hematological remission with a stable proviral load. To corroborate the observations of clinical remission, we used two sensitive molecular methods—high-throughput proviral integration site mapping and integration site-specific qPCR—to quantify the absolute and relative abundances of the malignant clone. The results of the two approaches agreed closely; the sensitivity of qPCR was greater, probably because the HTS quantifies the abundance of the large number of non-malignant HTLV-1-infected clones in addition to the malignant clone. By contrast, the qPCR method is clone-specific.

Whilst a rapid normalisation of the blood lymphocyte count was observed, the HTLV-1 PVL declined more slowly over the following months and the relative and absolute abundance of the presumed malignant clone remained high until 3.5 years after starting ZDV/IFN. Significantly, from 4.5 years onwards, without additional therapy, these diminished further culminating in a reduction of > 1000-fold in the absolute number of cells with this UIS. This reduction has been maintained for > 5 years without therapy. Although the mechanism of effect of ZDV/IFN-α has not been fully elucidated, the kinetics of this delayed reduction in the abundance of the malignant clone is consistent with the widely accepted hypothesis that ZDV/IFN-α does not exert direct cytotoxic activity or antiviral effects on leukemic cells at these doses. The anti-Tax and anti–HBZ CTL responses, measured by flow cytometry at diagnosis, during ZDV/IFN- α treatment and whilst in remission off therapy, were low-frequency and did not change significantly over time. As the tumour cells were type 2 defective, and did not express tax, we hypothesise that these responses are to non-malignant HTLV-1 infected cells (data not shown).

The use of ZDV/IFN-α has not only changed the management of patients with indolent subtypes of ATL, but has also raised a number of important clinical questions, relating to the optimum duration of treatment; whether patients can be regarded as cured by this approach; and whether biomarkers or other factors can be identified which might predict likely treatment responders. Earlier allogeneic bone marrow transplantation remains an option for those with a suboptimal response. ZDV/IFN-α is regarded by practitioners as a life-long treatment for ATL, but the treatment has significant side-effects and is often poorly tolerated.

In this patient, the integration site associated with the malignant clone remained detectable during clinical remission, albeit at low abundance. We suggest that this persistent clone might represent the parent clone from which the malignant population later arose. In non-malignant HTLV-1 infection, infected clones as identified by UIS can persist for many years, perhaps indefinitely, in stable equilibrium with the host immune response. ATL arises in high load HTLV-1 carriers (> 4%) [17, 18] and whilst most patients with ATL relapse with the dominant clone found at the initial presentation, several cases of clonal succession have been reported, with emergence of an infected clone carrying an HTLV-1 integration site and a TCRvB phenotype distinct from the malignant clone found at the initial presentation [19]. We, therefore, suggest that this patient’s current stable PVL ~ 10% reflects a return to his pre-morbid steady-state and that relapse of the previous malignant clone is now less likely than de novo transformation of a new clone.

This case demonstrates that stopping ZDV/IFN-α in chronic ATL can be associated with long clinical remission and, for the first time we have shown molecular evidence to support this clinical observation. We suggest that sensitive methods for detection of minimal residual disease (MRD) should now be applied in ATL clinical trials, to correlate clinical responses with MRD and to develop response-specific treatment protocols. In chronic ATL, application of these techniques could identify those patients who might benefit from alternative approaches such as early allogeneic bone marrow transplantation whilst sparing those who have a good prognosis from such an intensive approach.