Research ArticleCancer

Minimal Residual Disease Monitoring with High-Throughput Sequencing of T Cell Receptors in Cutaneous T Cell Lymphoma

See allHide authors and affiliations

Science Translational Medicine  04 Dec 2013:
Vol. 5, Issue 214, pp. 214ra171
DOI: 10.1126/scitranslmed.3007420


Mycosis fungoides (MF) and the leukemic presentation Sézary syndrome (SS) are clonal T cell lymphomas arising from the skin and are considered noncurable with standard therapies. To develop a specific and sensitive monitoring tool, we tested the ability of high-throughput sequencing (HTS) of T cell receptors (TCRB) to monitor minimal residual disease (MRD) after allogeneic hematopoietic cell transplantation. Genomic DNA was extracted from peripheral blood mononuclear cells (PBMCs) or skin samples. The rearranged TCRβ loci were amplified using Vβ- and Jβ-specific primers, followed by HTS, to generate up to 1,000,000 reads spanning the CDR3 region of individual cells. Malignant clones were identified in diagnostic samples in all cases by a dominant CDR3 sequence. Before transplant, four patients had circulating Sézary cells by the routine flow cytometry, which was confirmed by TCRB HTS. Although the flow cytometry found no detectable Sézary cells, malignant clones were detected by TCRB HTS in all other six cases. Five patients achieved “molecular remission” in blood between +30 and +540 days after transplant. Four of these patients also achieved molecular clearance in skin after transplant. Experiments using blood samples spiked with purified Sézary cells demonstrated that TCRB HTS can detect Sézary cells at the level of 1 in 50,000 PBMCs, which is more sensitive than standard diagnostics. We have thus demonstrated the utility of TCRB HTS to assess MRD with increased sensitivity and specificity compared to other current methodologies, and to monitor response to therapy in this MF/SS patient population.


Mycosis fungoides (MF) and the leukemic presentation Sézary syndrome (SS) are the most common types of cutaneous non-Hodgkin’s lymphoma (1, 2). Skin involvement by tumor cells in the form of patch, plaque, and tumor is more common in patients with MF, whereas diffuse erythrodermic change of most of the body surface is usually observed in SS patients (2). Extracutaneous involvement may occur in the lymph node, liver, lung, or peripheral blood with a significant level of circulating tumor cells (>1000 Sézary cells/μl) defining SS. The clinical, histologic, and immunophenotypic characteristics in patients with MF and SS have been well described, and the diagnosis is based on clinicopathologic and peripheral blood data correlation (3, 4). However, many patients are followed with nonspecific skin manifestation for several years before a diagnosis is made because of the lack of a tumor-specific diagnostic tool (5). The current diagnostic criteria rely heavily on clinical judgment along with consideration of immune histopathology, flow cytometry, and clonality demonstrated by population-averaged T cell receptor (TCR) polymerase chain reaction (PCR) (1, 4, 6, 7). The most specific test among these tools using standard PCR does not always identify a true clonal TCR gene rearrangement in MF/SS (6, 8, 9), and a presumed clonal TCR gene arrangement can be found in some cases of chronic benign inflammatory skin disease (8, 9) or in the peripheral blood in the absence of any malignant process (10). Other challenges in the management of these patients include accurate assessment and monitoring of minimal disease burden with therapy. It has been shown that increasing disease burden in peripheral blood (Sézary cell count) is correlated with a poor clinical outcome (5, 11, 12). However, the standard assessment tools often do not allow for accurate quantification. Because these monitoring tools, including multiparameter flow cytometry and TCR PCR, are not tumor-specific, their sensitivity of detection of minimal residual disease (MRD) is insufficient when dealing with cases with low tumor burden. This lack of sensitivity can be challenging for determining meaningful clinical remission. Recently, allogeneic hematopoietic cell transplantation has been shown to be an effective treatment that results in prolonged disease control in some advanced-stage MF/SS patients (1315). The monitoring of MRD becomes critical after an effective therapy such as allogeneic transplant to assess potential curative outcome. Therefore, developing a cost-effective and universally applicable assay with tumor specificity and high sensitivity to monitor residual disease is critical to meet these clinical needs.

High-throughput sequencing (HTS) is an emerging technology that has provided insight into the complexity of the immune repertoire by analysis of B cell receptor (BCR) and TCR gene rearrangement (16, 17). HTS of rearranged BCR gene has been used to monitor minimal disease status and B cell reconstitution in patients with chronic lymphocytic leukemia (CLL) after transplant (18). Other studies have used HTS of TCR to detect MRD after therapy in acute T lymphoblastic leukemia (19) or to assess the T cell reconstitution after allogeneic transplant (20). Here, we tested whether HTS of TCR can provide a specific and sensitive diagnostic tool to monitor MRD in this cutaneous T cell lymphoma population after allogeneic transplant. Besides testing for MRD in the blood after transplant, we also demonstrated the utility of TCR HTS in assessing MRD status in the skin.


Identification of malignant clone by HTS

Blood samples and selected skin biopsy samples from 10 patients with SS were collected as part of a prospective clinical trial. The complementarity-determining region 3 (CDR3) of TCRB was amplified and sequenced. Blood samples collected from eight patients with significant blood involvement were sequenced to identify the single dominant rearranged TCRB sequences representing the clonal neoplastic T cells. Archived formalin-fixed, paraffin-embedded skin biopsy samples were sequenced in the other two patients because there were no appropriate pretherapy blood samples available on these two cases. The frequency of these malignant clonal sequences represented between 18 and 92% of the total T cell repertoire (Table 1). To demonstrate that the clonal sequence remains the same in blood and skin, we sequenced the DNA extracted from an archived skin biopsy closer to the initial diagnosis in patient 1 (obtained 1.5 years before the blood sample). The identical dominant TCRB CDR3 sequence was found in the skin as in the later blood sample (representing 82% of the T cell repertoire in archived skin sample). The specific TCR Vβ family used by the clonal T cells had been identified using Vβ family–specific antibodies by flow cytometric analysis in six patients. It was identical to the one found by TCRB HTS in all six cases (Table 1). No suitable early blood samples could be tested by flow cytometry in two cases (patients 3 and 7). The flow cytometry–based TCR Vβ family detection kit failed to identify tumor-specific Vβ usage in patients 1 and 5.

Table 1. Characteristics of malignant table-wrap id=TCR clone.

Clonal TCRB CDR3 sequences were identified in all 10 patients. Nucleotide insertion (underlined) and D gene (italicized) regions are shown. The V and J gene family (number) or gene segment usage is indicated. The Vβ family usage identified by flow cytometric analysis is shown with corresponding V gene segment in parenthesis. n/a, not available.

View this table:

To evaluate the specificity of TCRB HTS for monitoring MRD, we determined whether the diagnostic malignant clonal sequence in any given patient was present within the TCR repertoire of any of the samples from any of the other patients or in blood samples from four healthy donors. Six reads (of a total of 6,675,236 TCRB reads in that sample or 0.00009%) of a CDR3 identical to the malignant clonal sequence of patient 4 were found in one posttransplant blood sample from patient 6 (table S2). This was the only example of a shared sequence (even at this very low level) among the 967 cross comparison performed. The specificity of TCRB HTS in identifying a malignant clonal sequence was thus 0.9989.

Sensitivity of detection of clonal sequence

To estimate the sensitivity of detection using TCRB HTS in clinical samples, we performed a spike-in experiment. Sézary cells were enriched from a whole-blood sample from patient 4 by flow cytometric sorting using a tumor-specific Vβ antibody. The postsorting sample showed 99.5% cell staining positive for tumor-specific Vβ 18. The enriched Sézary cells were then spiked into whole-blood samples from a healthy donor in triplicate at ratios of sorted tumor cells to total mononuclear cells from 1:100 to 1:1,000,000. DNA was extracted from these spiked samples in the same fashion as clinical samples, and sequenced as such. The malignant clonal sequence was found and represented 99.79% of the total T cell repertoire in the enriched sorted tumor sample. The malignant clonal sequence was detected in all triplicate samples between 1:100 and 1:50,000 [corresponding to a sensitivity of 1:150,000 total nuclear cells assuming that the mononuclear (lymphocyte/monocyte) cells make up about 30% of the total white blood cell count] (Fig. 1).

Fig. 1. Detection of malignant clonal sequence in spiked blood samples.

Malignant clonal sequence detected by TCR HTS was plotted as percentage of the total T cell repertoire in each sample. Each dot represents one sample. The horizontal bars represent the mean value of each spike-in level.

Monitoring of MRD

One goal of using TCRB HTS is to develop a sensitive, tumor-specific, and quantitative tool to monitor MRD after therapy. We looked at the blood samples collected before transplant and at different time points after transplant. At the time of preparative regimen for transplant, 9 of 10 patients had clinical evidence of active disease (table S1), including 4 patients with circulating Sézary cells found by standard multiparameter flow cytometry. The other six patients had no detectable Sézary cells by flow cytometry. However, having predetermined the index clonal sequences from the presentation blood or skin samples of these six patients, we could identify the presence of the malignant clones at 0.35, 0.33, 0.38, 0.03, 0.10, and 0.58% of the TCR sequences in these six cases. All except patient 2 achieved clinical complete response at day +90 after transplant. Patient 2 had mild persistent skin rash after transplant (table S1). The percentage of malignant clones decreased in all cases immediately after transplant. Five patients (1, 3, 5, 6, and 7) eventually achieved molecular clearance as measured by the HTS assay at days +270, +30, +30, +270, and +540 after transplant, respectively (Fig. 2 and table S2). The donor T cell chimerism in blood at the time of molecular remission was 97, 96, 46, 99, and 95% in these five patients, respectively (Fig. 2). In patient 7, the tumor clone was not detected from the day +30 blood sample but reappeared between days +60 and +360 at low frequency. Patients 8 and 9 had persistent but <0.05% tumor clones in their last blood samples. Patients 2 and 4 experienced clinical relapse in both skin and blood at days +180 and +360, respectively. The surge of tumor clones in the blood was evident in these two cases by TCRB HTS. For patient 2, the most recent analysis before clinical relapse was at day +60 (4 months before clinical relapse). For patient 4, the most recent analysis before clinical relapse was at day +180 (6 months before clinical relapse). There was no significant change in the MRD burden at these time points before clinical relapse (fig. S1). Patient 10 experienced clinical relapse in the skin and lymph nodes at day +480. The MRD burden in blood had significantly increased from a nadir of 0.034% at day +90 to 0.353% at day +270 (7 months before relapse), and to 1.632% at day +360 (4 months before relapse) (fig. S1). In contrast to patients who achieved molecular remission in blood, the highest donor T cell chimerism achieved in blood was 38, 23, 65, 30, and 98% in patients 2, 4, 8, 9, and 10, respectively. Both patients 2 and 4 lost their donor graft (donor T cell <5%) at the time of disease relapse, and no donor lymphocyte infusion (DLI) was given. Patient 10 is undergoing systemic therapy for disease control. None of the other seven patients showed sign of clinical relapse during their most recent follow-up.

Fig. 2. Residual disease status in the blood.

Malignant clonal sequence detected by TCR HTS was plotted as percentage of the total T cell repertoire at different time points. Open circle, detectable residual disease in blood; filled circle, molecular remission in blood; open square, diagnostic clone in skin. Imm Sup, duration and intensity of immunosuppression. CD3 ≥ 95%: solid bar represents the time period with full donor cell chimerism (donor T cells >95%); open bar represents donor T cells <95%.

Patient 1 developed acute graft-versus-host disease (GVHD) of the skin (grade 2) at day +36 and was treated with a short course of systemic steroid to resolution. All immunosuppressions were tapered off by 6 months after transplant in patient 1 as planned. Patient 10 developed acute GVHD of skin (grade 3) at day +87, which became chronic and required a prolonged course of immunosuppression. At the time of clinical relapse at day +480, patient 10 was receiving oral prednisone (5 mg/day), therapeutic tacrolimus, and extracorporeal photopheresis. Patient 3 developed extensive chronic GVHD involving the oral cavity, skin, and gut at day +150. This patient required prolonged systemic immunosuppression throughout the follow-up period. None of the other seven patients had either acute or chronic GVHD, and all immunosuppressions were tapered off by 6 months after transplant as planned (Fig. 2).

Posttransplant skin biopsy samples were available for TCRB HTS in 9 of the 10 patients. Skin samples from patients 4 and 10 were obtained at the time of relapse, showing a 66 and 63% malignant clone representation, respectively. Four patients (1, 3, 5, and 7) achieved molecular clearance measured by TCRB HTS in skin biopsy samples from previously involved areas (Table 2). However, MRD in skin was detected in the other three patients. The tumor clones contributed 0.29, 0.11, and 0.14% of the skin T cell repertoire in their most recent skin biopsy samples in these three cases.

Table 2. MRD in skin.

The percentage of the malignant clonal sequences in skin biopsy samples is shown.

View this table:

The most commonly used routine TCR PCR amplifies the rearranged TCR with a standard set of fluorochrome-labeled family-specific primers against the V, D, and J sections (BIOMED-2), followed by denaturing polyacrylamide gel or capillary sequencing polymer separation (21). The PCR products are detected by automated scanning (GeneScanning). The establishment of clonality is based on the appearance of a peak in the PCR products at a certain size. To show that multiple nonmalignant clones can give rise to a PCR product of the same size as the malignant clone in each sample, we analyzed the frequency and number of clones using the same TRBV and TRBJ, and with the same CDR3 length as the malignant clones (“mimic clones”). As shown in Table 3, mimic clones were found in all but 5 of the 60 posttransplant blood samples. The number of mimic clones in each sample varied significantly (median, 8; range, 1 to 89).

Table 3. The percentage of mimic clonal sequences.

The percentage of the malignant clonal sequences and the sum of percentage of the corresponding mimic clones in posttransplant blood samples are shown. The number of mimic clones in each sample represents the total number of different rearranged TCRB sequences with the same V and J family/gene segment usage and the same CDR3 length as the malignant clone in specific patient. Flow cytometry: Lymphocyte %, lymphocyte percentage of total nucleated cells; T cell %, T cell percentage of total nucleated cells; CD4+ CD26 %, percentage of CD4+ T cells that stained negative for CD26; CD4+ CD7 %, percentage of CD4+ T cells that stained negative for CD7.

View this table:

Multiparameter flow cytometry was used to detect the presence of residual Sézary cells in all except 2 of 62 posttransplant blood samples. CD4+ CD26 or CD4+ CD7 is the most commonly used phenotype for designation of the malignant population in MF or SS (22). Because a small proportion of normal T cell and reactive T cells have the CD4+ CD26 or CD4+ CD7 phenotype, the data suggest that a normal value for CD4+ CD26 or CD4+ CD7 cells by flow cytometry is lower than 15% (22, 23). Of the 37 MRD-positive samples by TCR HTS, 32 samples had <15% of CD4+ CD26 or CD4+ CD7 populations, whereas 5 samples had >15% of CD4+ CD26 or CD4+ CD7 subsets in the 23 MRD-negative samples.


The specificity of a test is critical for disease-monitoring tools especially in the context of MRD. Currently, the most specific confirmatory test for this disease is TCR PCR. The standard practice is to define a malignant clone by the appearance of the same monoclonal peak in at least two clinical samples from the same patient, either repeat samplings from the same compartment or samples from different compartments (that is, skin and blood) (1, 21, 24). Once the clonality is established by routine TCR PCR, only the relevant clonal peak is monitored in the subsequent follow-up study. However, even this “monoclonal” peak may not be tumor-specific. Indeed, multiple nonmalignant T cell clones with the same TRBV, TRBJ usage, and CDR3 length can give rise to similar PCR products (mimic clones), which cannot be differentiated by routine TCR PCR (Table 3).

In contrast, we have demonstrated the specificity of TCRB HTS in detecting MF/SS cells by the following. First, we have found only one malignant clonal sequence at very low frequency in 1 of 967 cross comparisons performed, which gave a very high estimated specificity (0.9989). With coincidental finding of cross-patient sequences being very rare, TCRB HTS proves to be an excellent tool in monitoring MRD after therapy. This result is consistent with a recent report on the low “false-positive” rate in detecting malignant T cell acute lymphoblastic leukemia (ALL) clones using TCRB HTS (19). Second, we have shown that the identical dominant malignant clonal sequence was found in an archived skin biopsy sample and blood sample that were obtained more than 1 year apart in patient 1. We have also found the identical dominant sequences in both the skin and blood at the time of clinical relapse, which was more than 1 year after transplant in patients 4 and 10. This finding is consistent with the notion that the sequence being followed in the individual patient is indeed the MRD of the malignant clone instead of a nonmalignant lymphoproliferative reaction. Third, in the six cases in which we have found the tumor-specific Vβ usage by family-specific antibodies to detect the expressed protein, the TCRB HTS identified exactly the same rearranged TRBV gene family.

Although identifying the dominant malignant clonal sequence by TCRB HTS is straightforward in suitable clinical samples, the lack of samples close to diagnosis when the malignant cells are abundant can be a limitation. Here, we used two archived formalin-fixed, paraffin-embedded skin samples to identify the malignant clonal sequence. Our ability to work with archived biopsy samples offers the opportunity to perform retrospective studies using TCRB HTS.

The current standard for MRD monitoring in MF/SS relies on multiparameter flow cytometry and TCR PCR to estimate the tumor burden and confirm the relevant clonality. The usual practice is to use flow cytometry to estimate the tumor burden and TCR PCR to confirm the relevant clonality (1). The sensitivity of flow cytometry suffers in cases with low tumor burden because of the lack of a tumor-specific marker. One way to improve both the specificity and sensitivity of flow cytometry is to use antibodies against tumor-specific Vβ family. However, the current commercially available panel of anti-Vβ family antibodies only covers 70% of the potential Vβ usage. In our hands, we were able to identify the tumor-specific Vβ usage in about 50% of the blood samples from SS patients, which was consistent with another report (25). Along with the CD4+ CD26 or CD4+ CD7 phenotype, the additional tumor-specific Vβ staining should be able to increase the sensitivity of detecting residual disease by multiparameter flow cytometry.

The routine TCR PCR is neither tumor-specific nor very sensitive. To improve its detecting power, allele-specific oligonucleotide PCR (ASO-PCR) (2628) and combining cell sorting of “suspect” tumor cells with routine TCR PCR have been attempted (29). However, the ASO-PCR required the cloning and sequencing of the CDR3 region of TCR of every individual case. Although ASO-PCR can be a very sensitive quantitative assessment using real-time PCR technology, this process requires weeks to perform and is not routinely available. Here, we showed that HTS of TCRB provides a universally applicable disease-monitoring tool with high specificity and sensitivity. Indeed, TCRB HTS can consistently detect 1 tumor cell in 50,000 peripheral blood mononuclear cells (PBMCs) (or in about 150,000 leukocytes equivalent) (Fig. 1). This level of sensitivity has exceeded the standard TCR PCR by at least two logs, which has a sensitivity of 1 in 50 to 100 cells (21). It is similar to ASO-PCR, which can detect tumor cells in concentrations as low as 0.01 to 0.001% (1 tumor cell in 10,000 to 100,000 healthy cells) (28). Increasing the depth of TCRB HTS can increase the sensitivity of detecting a malignant clone. As of the writing of this manuscript in our current protocols, we set the depth of sequencing to yield fivefold coverage of 200,000 genomes. Compared to ASO-PCR, the advantage of using HTS of TCRB is the elimination of custom-made agents and a short turnaround time of 1 week. The short turnaround time makes it feasible for clinical decision-making.

TCR HTS has provided a tool to study other important aspects of allogeneic transplant. First, the delayed clearance of the malignant clone observed in three patients suggested a slow but persistent graft-versus-lymphoma effect in these cases. This is similar to the delayed molecular remission 1 year after transplant in CLL patients who have undergone nonmyeloablative allogeneic transplantation (18). The timing of achieving molecular remission in our study was not entirely associated with full donor T cell chimerism, because patients 1 and 7 achieved full donor T cell chimerism (>95%) 8 and 12 months before molecular clearance of the malignant clone, whereas patient 5 achieved molecular remission at donor T cell chimerism of 46% (Fig. 2). On the other hand, four of the five patients who never achieved molecular remission had low donor T cell chimerism. Second, the persistence of malignant clone in the skin 9 months after achieving molecular remission in the blood in one patient (patient 6) implied a different kinetics of graft-versus-lymphoma effect in these two compartments. Third, “T cell repertoires” are generated by TCR HTS on every sample tested for the purpose of MRD monitoring. We can potentially determine the degree of T cell immune reconstitution at different time points after transplant by analyzing the size and diversity of the CDR3 sequences. The recent report on quantifying the T cell repertoire after allogeneic transplant has shown the possibility of such analysis (20).

Detection of progression of MRD by HTS can potentially lead to early intervention. Preemptive therapeutic intervention such as reduction of immunosuppression, DLI, or additional immunotherapy or biological therapy may be applied to delay or prevent clinical relapse. Here, MRD progression was not demonstrated by HTS before clinical relapse in two of the patients (patients 2 and 4). However, the MRD assessment was performed 4 months before relapse in one case and 6 months before relapse in the other without additional interval MRD monitoring. The optimal frequency of MRD monitoring has not been established in this population. The third relapsed patient (patient 10) did show MRD progression in blood as early as 7 months before clinical relapse in the skin and lymph nodes (fig. S1). Reduction of immunosuppression or even low-dose DLI might have been considered before clinical relapse in this patient. This observation is consistent with a previous report on the predictive value of MRD progression determined by HTS of immunoglobulin H in relapse in CLL patients (30). Although the concept of predicting clinical relapse by monitoring of MRD with HTS is an attractive one, additional studies with a larger patient population are needed to confirm this finding. It is also unclear whether there is a threshold for MRD to become clinically relevant. One of the limitations of this report is that the sensitivity and specificity of detecting MRD can be altered by changing the depth of HTS. The depth of sequencing is expected to increase when HTS becomes widely available and can be applied more routinely in the near future.

HTS of TCRB can aid in the clinical management after allogeneic transplant in this MF/SS patient population. Skin rash after allogeneic transplant can represent drug toxicity, GVHD, or disease progression in patients with MF or SS. Commonly, the diagnosis is based on clinical judgment and pathological review of skin biopsy samples, which can be quite challenging because of the similarities in the histologic and clinical features. The management is vastly different depending on the cause of the skin eruption. In the case of GVHD, more immunosuppression is needed, whereas less immunosuppression is desirable in the case of active MF or SS. Having the ability to monitor the MRD in a quantitative way using TCRB HTS is an important tool in managing these cases. On the other hand, the interpretation of MRD in skin by HTS can be limited by the sampling variation because molecular remission at one biopsy site may not represent a global remission. In addition to monitoring the MRD, HTS of TCRB can potentially identify a TCR repertoire signature in the cases of skin GVHD. Recently, using HTS technology to track GVHD-related TCR signature has been successfully applied in patients with gut GVHD (31). The analysis of skin TCR repertoire using HTS of TCRB in patients with skin GVHD may also be informative.

Achieving molecular remission has been shown to be associated with better clinical outcome in other hematological malignancies. In one report, molecular remission in blood at 1 year after allogeneic transplant was associated with a reduced risk of clinical relapse in CLL patients using ASO-PCR to detect MRD at the level of 1 in 10,000 leukocytes (10−4) (32). In another report, molecular response defined by BCR PCR was shown to be an independent predictor for better clinical outcome in pediatric B cell ALL (33). Recently, this association between molecular remission after allogeneic transplant and prolonged disease control has been confirmed in CLL patients using a more sensitive BCR HTS to detect MRD at the level of 1 in 1,000,000 leukocytes (10−6) (30). For patients with MF and SS, tumor detection and burden in the blood have significant prognostic impact (11, 12). Several reports have suggested that there was a negative prognostic effect associated with minimal disease burden detected only by TCR PCR (12, 34). With better specificity and sensitivity, TCRB HTS may be the superior tool to confirm this adverse outcome of minimal disease status. The impact of achieving molecular remission determined by either standard TCR PCR or high-sensitivity TCR HTS in MF/SS patients has not been a high priority, mostly because of the lack of therapies with lasting clinical remission or curative potential in this patient population. Given the successful clinical experience with allogeneic transplant (1315), monitoring MRD and determination of clinically meaningful molecular remission have become critical in managing these patients after transplant.


Study design

Peripheral blood and skin samples were from patients who were enrolled in a clinical trial at Stanford University School of Medicine ( identifier: NCT00896493) using nonmyeloablative allogeneic hematopoietic cell transplantation in advanced-stage MF and SS. The patients received a preparative regimen using total skin electron beam therapy, total lymphoid irradiation, and anti-thymocyte globulin, followed by allogeneic hematopoietic cell infusion. They also received cyclosporine and mycophenolate mofetil as GVHD prophylaxis. Blood samples were obtained before transplant and at different time points (days +30, +60, +90, +120, +180, +270, +360, +540, and +730) after transplant as part of a prospective study for MRD monitoring. Skin biopsy samples were obtained before transplant and at +90 days after transplant, and additional time points depending on the clinical status. This report included the first 10 SS patients in this clinical study. They were highly treated with a median of five previous systemic treatments. All patients received systemic treatment immediately before transplant in an attempt to reduce the disease burden. All except one patient had active disease in at least one of the three compartments—skin, blood, and lymph node—at the time of preparative regimen (table S1). This study was conducted according to an institutional review board–approved protocol, and informed consent was obtained from all patients.

Flow cytometric analysis and pathology review

Multiparameter flow cytometry was performed on whole blood obtained from consented patients through venipuncture. Briefly, the peripheral blood leukocytes were stained for CD45, CD3, CD4, CD8, CD19, CD7, and CD26. Abnormal Sézary cell population was defined as dimCD3+, dimCD4+ T cells with diminished or absent expression of CD26 and/or CD7 (22, 35, 36). The pathologist generates a hematopathology report with an interpretation of flow cytometric findings and description of percentage and absolute number of Sézary cells. To identify the tumor-specific TCR Vβ usage, we used a panel of antibodies specific to different Vβ families to stain circulating Sézary cells (IOTest Beta Mark TCR V Kit, Beckman Coulter).

Spike-in experiment

Antibodies against tumor-specific Vβ 18 were used for the enrichment of Sézary cells from an early blood sample of patient 4 using flow cytometric cell sorter. The enriched Sézary cells were then spiked into whole-blood samples from a healthy donor in triplicates in a series dilution manner. The Sézary cell–to–normal mononuclear cell ratios ranged from 1:100 to 1:1,000,000. DNA was extracted from the spiked samples, and DNA corresponding to 200,000 genomes was used for sequencing.


Genomic DNA was prepared from PBMCs, frozen 4-mm skin punch biopsies, or formalin-fixed, paraffin-embedded tissue with a Qiagen DNA extraction kit. DNA corresponding to about 200,000 genomes was used for sequencing. A multiplex PCR system was used to amplify rearranged TCRβ loci. A set of forward primers, each specific to a functional TCR Vβ segment, and reverse primers, each specific to a TCR Jβ segment, were used in solid-phase PCR (Illumina GA2 system or HiSeq platform) to generate reads that cover the entire CDR3 region (17). Raw sequence data were processed to remove errors. The distinct TCRB CDR3 sequences were plotted on the basis of their frequency of appearance to generate the TCR “repertoire.” The dominant CDR3 sequence of the tumor cells (malignant clonal sequence) was determined from the diagnostic samples.

Statistical analysis

Calculations of the specificity of the test were based on the following formula:

Embedded Image


Fig. S1. MRD progression at low tumor burden.

Table S1. Clinical characteristics of the patients.

Table S2. The percentage of malignant clonal sequence in all blood and skin samples tested.


  1. Funding: This work was supported by the Haas Family Fund, and Albert Yu and Mary Beckmann Foundation. W.-K.W. is the recipient of a Stanford University Cancer Center Developmental Cancer Research Award and Stanford Institute for Immunology, Transplantation and Infection Seed Grant Award. Author contributions: W.-K.W. designed the study, performed the experiments, collected clinical samples and data, interpreted the results, and wrote the manuscript. R.A. performed flow cytometric analysis and cell sorting. S.A. designed the study and interpreted the results. C.D. performed the HTS and interpreted the results. R.H. designed the study and interpreted the results. Y.H.K. designed the study, interpreted the results, and co-wrote the manuscript. Competing interests: C.D. has employment and equity with Adaptive Biotechnologies. The other authors declare that they have no competing interests.

Stay Connected to Science Translational Medicine

Navigate This Article