Real-world treatment patterns, healthcare resource utilization and costs among patients with peripheral T-cell lymphoma

Peripheral T-cell lymphomas (PTCL) are a rare and aggressive form of non-Hodgkin's lymphoma (NHL) that affect mature T- and natural killer cells. In the USA, the incidence of NHL is estimated to be around 80,000 cases per year and PTCL only accounts for up to 10% of all NHL cases [1–3]. PTCL is a heterogeneous disease that can be characterized into more than 25 specific subtypes based on clinically distinct features [4]. Accounting for more than 88% of PTCL cases in the USA, the most common subtypes include PTCL not otherwise specified (PTCL-NOS), angioimmunoblastic T-cell lymphoma (AITL), ALK-positive or negative anaplastic large cell lymphoma (ALCL), adult T-cell leukemia/lymphoma (ATLL) and natural killer/T-cell lymphoma (NKTCL) [5,6].

Given the clinical heterogeneity and aggressive nature of PTCL, its diagnosis and treatment can be challenging. Often diagnosed at an advanced stage, PTCL patients have a poor prognosis, with an estimated 5-year survival below 50% across most subtypes [7]. Moreover, relapse after first line of treatment (1L) is common and can be subtype dependent [3]. According to National Comprehensive Cancer Network (NCCN) guidelines, most PTCL subtypes are treated in the first line (1L) with an anthracycline-based chemotherapy combination, such as cyclophosphamide, doxorubicin, vincristine and prednisone (CHOP) or other CHOP-like regimens, with or without radiation. Cyclophosphamide, doxorubicin, vincristine, etoposide and prednisone (CHOEP) or CHOEP-like regimens are also recommended in some subtypes of PTCL [8]. Novel targeted therapies including antibody–drug conjugates such as brentuximab vedotin, given as monotherapy or in combination with an anthracycline-based combination regimen, are also being used in the 1L to treat PTCL [9]. Second-line (2L) treatment of patients with refractory or relapsed PTCL include treatment with single agents such as brentuximab vedotin, belinostat, pralatrexate and romidepsin. Additional treatment options include combination chemotherapy or palliative radiation therapy [8]. Follow-up consolidation treatment may be administered after patients have a complete or partial response to 1L or 2L therapy and may include clinical trials or treatment with high-dose chemotherapy followed by autologous or allogeneic stem cell transplant [8,10].

Due to the rare and heterogeneous nature of PTCL and its poor prognosis, there is limited understanding of the PTCL patient population and its real-world treatment patterns. Previously published retrospective studies have evaluated the burden of illness, treatment patterns and economic burden among newly diagnosed patients with PTCL in the commercial and Medicare populations, but evidence on treatments by lines of therapy (LOT) and associated economic burden is limited [11–14]. Findings from a recent systematic review of 55 clinical trials and real-world studies among PTCL patients reported that most of the published studies focused on CHOP or CHOP-like treatment patterns and their clinical outcomes [5]. The review also highlighted the lack of real-world data on the economic burden associated with PTCL treatments. To that end, only a few studies in the USA assessed healthcare resource utilization (HRU) and/or healthcare costs among PTCL patients [11–14]. A recent study including the Medicare PTCL population reported a mean adjusted all-cause cost of US$5885 per patient per month (PPPM) and mean disease-related costs of US$2093 PPPM during a median follow-up of 2 years [13]. In a US administrative claims database study, PTCL patients had higher monthly hospitalization, pharmacy utilization, emergency room visits and mean cost per patient compared with non-PTCL controls [14]. To that end, this retrospective study aims to describe real-world patient characteristics, treatment patterns, HRU and associated costs of patients with PTCL, overall and by LOT, based on recent data from a large US healthcare claims database.

Materials & methods

Study design & data source

This retrospective cohort study leveraged the IQVIA PharMetrics^® Plus patient-level claims database from 1 April 2011 to 30 November 2021 to identify patients with PTCL and their systemic PTCL treatments. IQVIA PharMetrics Plus is a health plan claims database comprised of fully adjudicated medical and pharmacy claims for more than 210 million unique enrollees. It contains a longitudinal view of inpatient and outpatient services, prescription and office/outpatient-administered drugs, costs and detailed enrollment information. The current study used de-identified patient records and does not involve the collection, use or transmittal of individually identifiable data, thus institutional review board approval was not necessary to conduct this study.

Study population & period

Adult patients (≥18 years) with ≥1 inpatient or ≥2 outpatient diagnoses (≥30 days apart) for PTCL (International Classification of Diseases, Ninth Revision of Clinical Modification [ICD-9-CM] and ICD-10-CM; Appendix 1) between 1 October 2011 and 31 October 2021 were identified. The date of the first claim of PTCL diagnosis was the index diagnosis date (overall PTCL cohort). Patients were required to have ≥6 months of continuous enrollment before (i.e., baseline) and ≥1 month of continuous enrollment after (i.e., follow-up) the index diagnosis date. Patients were also required to be without a non-PTCL primary cancer diagnosis within ±30 days of the index diagnosis date. Patients with evidence of ≥1 systemic treatment for PTCL on or after the index diagnosis date in the overall cohort were identified as the treated cohort; the date of first PTCL treatment claim during the study period was the index treatment date. Systemic treatments were identified using relevant Healthcare Common Procedure Coding System (HCPCS) codes and National Drug Codes (NDC) based upon the NCCN guidelines for PTCL treatment [8]. Patients in the treated cohort were further required to have ≥6 months of continuous enrollment before (i.e., baseline) and ≥1 month of continuous enrollment after (i.e., follow-up) the index treatment date. Patients were excluded if they had evidence of systemic PTCL treatment during the fixed 6-month baseline period. Patients in the treated cohort were further categorized into three mutually exclusive subcohorts based on the total number of LOTs observed during follow-up: 1LOT, 2LOT or ≥3LOT.

Treatment regimens for PTCL observed during follow-up (1st regimen, 2nd regimen, 3rd regimen or greater) among the treated cohort and LOT subcohorts were identified. A LOT regimen was defined as a segment of PTCL-specific treatment and could be a monotherapy or combination therapy regimen. The use of only one medication without any other agents observed within a 30-day window after the initial medication dispensing was considered a monotherapy regimen. If any other agent was initiated within the 30-day window after the initial medication claim, the regimen was considered a combination regimen. Among combination regimens, the CHOP-like regimen included combinations with cyclophosphamide, doxorubicin and vincristine while allowing for different corticosteroids (e.g., prednisone, methylprednisolone, dexamethasone) within the regimen. Similarly, CHOEP-like included combinations containing etoposide, cyclophosphamide, doxorubicin and vincristine along with any corticosteroid. The duration of a LOT was defined as the time from the LOT initiation to the LOT end date. LOT end date was based on the earliest of the following events:

• Discontinuation (gap in treatment of ≥60 days);

• Line advance (defined as a switch [e.g., switched from their index treatment to another PTCL treatment ≥30 days after LOT index treatment initiation] or augmentation [defined as evidence of addition of a non-index LOT treatment ≥30 days after LOT index treatment initiation]);

• End of continuous enrollment;

• End of study period.

Study measures

All study cohorts were described in terms of baseline demographic and clinical characteristics. All-cause and PTCL-related HRU and associated costs PPPM were assessed in the treated cohort and LOT subcohorts. PTCL-related HRU was defined as any service/claim with PTCL diagnosis code in any position. PTCL-related pharmacy claims included medications indicated for PTCL. All costs were reported in 2021 US$ based on the Consumer Price Index of Medical care for all Urban Consumers [15] and reported as PPPM.

Statistical analysis

Median duration of treatment and 95% CI were estimated using the Kaplan–Meier (KM) analysis for each LOT subcohort by the top treatment regimens observed. For the KM method, the end of treatment regimen was the earliest of the following: end of LOT; end of continuous enrollment; or end of study period. All analyses were performed using SAS version 9.4 (SAS Institute Inc., NC, USA).

Results

Patient characteristics

Among 900 patients meeting the selection criteria in the overall PTCL cohort, a total of 189 patients were selected into the treated cohort after applying the additional selection criteria. These 189 patients were further categorized into 1LOT (n = 117), 2LOT (n = 41) and ≥3LOT (n = 31) subcohorts (Figure 1). Nearly 35% (n = 312) of the 900 PTCL patients in the overall cohort were treated with steroids only and 39.0% (n = 351) did not have evidence of any treatments (including steroids; Figure 1); demographic and baseline clinical characteristics of these patients can be found in Supplementary Table 1.

The mean age of the overall PTCL cohort and treated cohort was 53.5 years (standard deviation [SD]: 15.5) and 54.6 years (SD: 15.1), with a majority being (53.6 and 59.3%, respectively) ≥55 years old. The overall PTCL and treated cohorts were predominantly male (57.7 and 62.4%) and mostly resided in the Southern (33.6 and 37.0%) or Midwest (26.7 and 33.9%) regions of the USA. The majority of patients in both cohorts were commercially insured (62.9 and 65.6%) and had a Preferred Provider Organization (PPO) plan type (70.6 and 72.0%; Table 1). ALCL and PTCL-NOS were the two most common PTCL subtypes observed in the overall PTCL cohort (40.9 and 48.0%) as well as the treated cohort (42.9 and 38.6%). The proportion of patients with PTCL-NOS subtype was higher in the overall cohort than the treated cohort (48.0 vs 38.6%). Most patients had a Charlson Comorbidity Index (CCI) score (excluding cancer) of 0 in the overall cohort (65.9%) as well as the treated cohort (54.0%). However, a higher proportion of patients in the treated cohort had a CCI score of 2 or higher (28.0 vs 17.7%). Similar comorbidities were observed to be prevalent in the overall PTCL and treated cohorts, including hypertension (28.6 and 36.5%), mild-to-moderate diabetes (13.7 and 15.3%) and chronic pulmonary disease (11.7 and 17.5%). During the 6-month baseline period, the mean all-cause total healthcare cost PPPM was $3815 (SD: US$13,225) for patients in the overall cohort and US$8040 (SD: $11,941) for the treated cohort.

For the LOT subcohorts, the mean age of patients ranged between 54.3 years (SD: 16.3) to 55.6 years (SD: 14.3), increasing with each LOT subcohort (Table 1). All the LOT subcohorts had a male majority population, with the highest proportion observed in the 1LOT subcohort (68.4%) and most patients in the LOT subcohorts were commercially insured with a PPO plan type. Specifically, 50.4% of the 1LOT subcohort had the ALCL subtype, while PTCL-NOS was more common than ALCL in the 2LOT (53.7 vs 24.4%) and ≥3LOT cohorts (41.9 vs 38.7%; Table 2). The mean CCI score was the lowest in ≥3LOT subcohort (0.7 [SD:1.1]) and was similar for 1LOT (1.1 [SD: 1.6]) and 2LOT (1.1 [SD: 1.5]) subcohorts. Similar to the overall cohorts, hypertension (40.2, 29.3 and 32.3%), chronic pulmonary disease (20.5, 9.8 and 16.1%) and mild-to-moderate diabetes (16.2, 14.6 and 12.9%) were the most prevalent comorbidities. For patients in the LOT subcohorts, steroid treatment before initiation of systemic therapy was common at 57.3, 51.2 and 41.9% for 1LOT, 2LOT and ≥3LOT cohorts, respectively. The most commonly utilized steroid was prednisone, followed by dexamethasone and methylprednisolone. Mean baseline all-cause total healthcare costs were highest for the 2LOT subcohort (US$9468 [SD: $12,878]), followed by 1LOT (US$8218 [SD: $12,704]) and ≥3LOT (US$5482 [SD: $6223]) subcohorts.

In an exploratory analysis of the overall cohort, it was observed that during follow-up 34.7% (n = 312) of patients were treated with steroid monotherapy only (no systemic PTCL treatment) and 39.0% (n = 351) received no treatment. The patients with steroid-only treatment had a similar mean age than the no-treatment patients (54.7 [SD: 14.3] vs 51.6 [SD: 16.6]) and more comorbidities, including higher proportions of chronic pulmonary disease (16.3 vs 9.1%). The mean baseline costs were higher among patients treated with steroid monotherapy (US$4108 [SD: $18,334] vs $3388 [SD: $10,730]).

Treatment regimens

Approximately 62% of patients in the treated cohort had 1LOT, 21.7% had 2LOT and 7.4% had 3LOT; the remaining 9.0% of patients had ≥4LOT regimens observed during follow-up (Table 3). Monotherapy regimens were more commonly observed than combination therapies for PTCL in the 2LOT and ≥3LOT, while combination therapies were used more frequently in the 1LOT. For monotherapy, the most common 1st treatment regimens were brentuximab vedotin (13.8%), methotrexate (7.4%) and romidepsin (4.2%). For combination therapies, the most common regimens included CHOP and CHOP-like (25.9%), CHOEP and CHOEP-like (20.1%), and a combination of brentuximab, cyclophosphamide and doxorubicin (5.8%). These therapies remained most common 1st regimens for individual LOT subcohorts as well. In the 2nd treatment regimen (among those patients who had at least 2LOT), brentuximab (16.7%), romidepsin (9.7%) and methotrexate (8.3%) remained the most common monotherapy regimens. Only a small number of patients were using combination therapy as the 2nd treatment regimen. In the 3rd treatment regimen, brentuximab (19.4%), romidepsin (16.1%) and pralatrexate (12.9%) were observed as the top three monotherapy regimens for PTCL.

Nearly 62% of patients in the treated cohort discontinued their index LOT treatment: 67.5% discontinued in the 1LOT, 53.7% in the 2LOT and 51.6% in the ≥3LOT. By study definitions, since patients in the 1LOT subcohort had only one treatment regimen, switching between regimens was not possible. For the 2LOT and ≥3LOT subcohorts, 80.5 and 87.1% of patients switched from their index regimen to another PTCL treatment regimen, respectively. The median time on treatment for the top five 1LOT treatment regimens was 133 days (95% CI: 118–139) for CHOP and CHOP-like, 137 days (95% CI: 135–142) for CHOEP and CHOEP-like, 133 days (95% CI: 98–161) for brentuximab vedotin, 130 days (95% CI: 34–808) for methotrexate and 133 days (95% CI: 51–not estimable [NE]) for brentuximab vedotin, cyclophosphamide and doxorubicin combination (Figure 2A). For the most frequently observed 2LOT treatment regimens, the median time on treatment for brentuximab vedotin was 133 days (95% CI: 49–201) and romidepsin was 315 days (95% CI: 43–NE). The median time on treatment for methotrexate as the 2LOT regimen was 183 days (95% CI: 28–NE; Figure 2B). The median time on treatment for the top three treatment regimens in the ≥3LOT subcohort was 196 days (95% CI: 28–NE) for brentuximab vedotin, 207 days (95% CI: 42–NE) for romidepsin, and it was not estimable for pralatrexate (Figure 2C).

Approximately 17% of patients in the treated cohort had at least one stem cell transplant, with the highest proportion observed among ≥3LOT patients (22.6%), followed by 1LOT (16.2%) and 2LOT patients (14.6%). Radiation therapy was observed in 13.2% of the treated cohort, with an increasing proportion in LOT cohorts from 9.4% in 1LOT, 9.8% in 2LOT and 32.3% in ≥3LOT subcohort.

HRU & costs

The treated cohort had high all-cause HRU in medical services across all care settings and pharmacy services, with nearly 60% of the treated cohort having ≥1 hospitalization. Among patients with ≥1 hospitalization in the treated cohort, the mean length of stay per hospitalization was 10.1 (SD: 11.7) days. Among the LOT subcohorts, the proportion of patients with ≥1 hospitalization increased by increasing LOT, with the highest among the ≥3LOT subcohort at 77.4%, followed by 2LOT at 73.2% and 1LOT at 49.6% (Figure 3A). Emergency room (ER) visits also had an increasing trend with each successive LOT, with 40.2, 53.7 and 71.0% of patients having ≥1 ER visit in the 1LOT, 2LOT and ≥3LOT, respectively. Among nearly all patients with at least one physician office visit, the mean number of visits PPPM was 2.8 (SD: 3.5) for the treated cohort, with minor differences observed for the LOT subcohorts (1LOT: 2.8, 2LOT: 2.5 and ≥3LOT: 3.1 visits PPPM). The mean number of prescriptions PPPM was 8.0 (SD: 7.6) in the treated cohort, with an increase in prescriptions observed with increasing LOT (1LOT: 7.6, 2LOT: 8.1 and ≥3LOT: 9.6 prescriptions PPPM).

PTCL-related hospitalization was observed in 43.4% of patients in the treated cohort, with the mean length of stay per hospitalization of 10.6 days (SD: 12.7). Similar to all-cause HRU, the proportion of patients with at least one hospitalization was higher in the 2LOT and ≥3LOT subcohorts compared with the 1LOT cohort, respectively (Figure 3B). For the LOT subcohorts, the proportion of patients with ER visits increased with increasing LOTs. Physician office visits and outpatient labs were high in the treated cohorts as well as LOT subcohorts. The mean number of PTCL-related prescriptions PPPM for the treated cohort was 3.6 and was highest for the ≥3LOT subcohort (4.1; SD: 4.6). Chemotherapy administration was observed to be present in nearly 80% of the treated cohort, with minor differences observed in the LOT subcohorts.

The mean all-cause total healthcare cost PPPM for the treated cohort was US$24,924 (SD: $29,788); three times higher than mean baseline costs (US$8040). This trend persisted for each LOT subcohort. Prescription drugs and hospitalizations were the major cost drivers in all LOT subcohorts, accounting for approximately two-thirds of total costs (Figure 4A). Among the three LOT subcohorts, mean cost for prescription drugs was highest in the ≥3LOT subcohort (US$15,356 [SD: $14,261]), whereas mean costs for hospitalization was highest in the 2LOT subcohort (US$11,848 [SD: $26,059]). In a sensitivity analysis among a subgroup of the treated cohort with at least 12 months of follow-up (n = 114), the mean total healthcare costs PPPM was US$14,073 (SD: $14,315). The mean total healthcare costs for the LOT subcohorts in the sensitivity analysis were US$9178 (SD: $11,354), $15,555 (SD: $12,684) and $22,997 (SD: $17,209) for the 1LOT, 2LOT and ≥3LOT subcohorts, respectively.

The mean total PTCL-related healthcare cost PPPM for the treated cohort was US$17,251 (SD: $26,340), accounting for 69.2% of the all-cause total healthcare costs. Mean total PTCL-related healthcare costs were highest among the 2LOT subcohort (US$19,824 [SD: $32,387]), followed by ≥3LOT subcohort (US$17,438 [SD: $15,960]). Among the resource categories, hospitalizations, prescription drugs and chemotherapy administration accounted for more than 80% of the total PTCL-related healthcare costs in the LOT subcohorts (Figure 4B). Similar to the all-cause healthcare costs, the mean costs for prescription drugs were highest in the ≥3LOT subcohort (US$9967 [SD: $11,478]) and mean costs for hospitalizations were highest in the 2LOT sub-cohort (US$8991 [SD: $25,424]). In the separate sensitivity analysis among the treated cohort with at least 12 months of follow-up (n = 114), the mean total PTCL-related healthcare costs PPPM was US$9698 (SD: $12,724). The mean total PTCL-related healthcare costs for the LOT subcohorts in the sensitivity analysis were US$6156 (SD: $9932), $9841 (SD: $11,710) and $17,155 (SD: $15,934) for the 1LOT, 2LOT and ≥3LOT subcohorts, respectively.

Discussion

This retrospective database study provides a contemporary analysis of the real-world treatment patterns and economic burden among PTCL patients in the USA. To our knowledge, this is the first study to describe PTCL treatment regimens by LOT and estimate the HRU and associated costs using a US commercial claims database. Among diagnosed PTCL patients, only 21% had evidence of systemic treatment for PTCL during the follow-up period, with 62% having 1LOT, 22% having 2LOT and 16% having ≥3LOTs. The mean age of the overall cohort and the treated cohort was approximately 54 years, similar to previously published claims-based studies among PTCL patients [5,14]. PTCL is diagnosed more frequently in males compared with females, and this sex imbalance was observed in our study, with 57.7% males in our overall cohort [16]. However, the sex imbalance diminished by the time patients reached ≥3LOT. Consistent with published studies, the most common comorbidities in our study were hypertension, diabetes and chronic pulmonary disease, across all study cohorts [14,16–18].

It is noteworthy that in our study, over one-third of diagnosed PTCL patients were treated with only steroid treatment, particularly prednisone and dexamethasone. Moreover, nearly half of the patients with systemic PTCL treatment had prior steroid use, potentially for management of other comorbid conditions or PTCL symptoms, yet the reason behind high steroid use warrants further investigation. According to published clinical guidelines [8], steroid treatment is typically indicated for PTCL patients in combination with systemic PTCL treatment such as chemotherapy combinations. However, steroids can also be used as a maintenance therapy for patients undergoing stem cell transplants, radiation therapy, surgical intervention or those with relapsed/refractory PTCL. Steroids may also have been utilized to treat other comorbidities unrelated to PTCL, such as chronic pulmonary disease, liver disease and rheumatologic disease. For example, in exploratory analyses we found that patients with steroid monotherapy treatment had a higher prevalence of chronic pulmonary disease (16.3 vs 9.1%) compared with patients with no treatment. Our finding is in line with another retrospective claims database study among Medicare patients, where 17% of the PTCL patients were treated with steroids only [13]. Further investigation is warranted to understand steroid utilization within the treatment paradigm of PTCL.

The current study observed PTCL treatment patterns consistent with the published literature, suggesting CHOP or CHOP-like treatment regimens (29.1%) are the most common combination first-line therapies among PTCL patients [16,19,20]. Following CHOP and CHOP-like treatment, CHOEP or CHOEP-like regimens (18.8%) and brentuximab vedotin combination therapies (8.5%) were also observed in the 1LOT. This is in line with the recommendation on including etoposide for improved response and using brentuximab vedotin as a part of combination therapy [8,9]. Median treatment duration for 1LOT treatment regimens, including CHOP/CHOP-like and CHOEP/CHOEP-like, was slightly over 4 months and consistent with their established dosing durations of six treatment cycles [8]. Based on the PTCL treatment guidelines, most patients were expected to initiate a combination chemotherapy treatment in the 1L [8,21]. However, we observed a significant number of patients with monotherapy use in all LOTs, with brentuximab vedotin, romidepsin and methotrexate being the most commonly utilized treatments [22]. Brentuximab vedotin, given its recent approval by the US FDA for untreated systemic PTCL patients and its established efficacy with lower adverse event risk profile, was the most common monotherapy observed in all LOT subcohorts [22]. Moreover, half of the 1LOT subcohort had an ALCL subtype expressing CD30, making them better candidates for brentuximab-based treatment regimen [8,9,23]. A high proportion of patients with monotherapy and other treatment regimens (steroids-only treatment) were observed in our study. This was consistent with a recent study by Shah et al. which reported that nearly 71% of patients treated with any PTCL systemic regimen utilized a treatment that is not a recommended therapy as per NCCN guidelines [13].

Treatment regimen discontinuation and switching were high in all LOT subcohorts. Discontinuation was highest in 1LOT, and switching was highest in ≥3LOT. Since there are no previously published studies reporting a rate of discontinuation by LOT, it is not possible to compare our study with other published studies. However, evidence from the literature suggests that treatment discontinuation and switching most likely occurred due to disease progression and/or presence of adverse events, although reasons for discontinuation were not explored in this study [6,24]. It is possible that a small number of patients were considered to have discontinued the LOT due to end of continuous enrollment or end of study period; however, the median length of follow-up for each LOT subcohort allows for ample follow-up time for the majority of patients to be assessed for discontinuation based on the gap in treatment definition. In addition, a proportion of patients may reach partial or complete remission after systemic treatment, particularly after 1L therapy, leading to treatment discontinuation. Consistent with the published literature, only 3.7 and 13.2% of patients in the treated cohort were observed to have allogeneic or autologous cell stem cell transplants, respectively [10,25]. Several studies have reported positive outcomes with consolidated therapy (high-dose therapy followed by stem cell transplant) in a 1L setting [8,26,27]. In this study, the proportion of patients with stem cell transplants were lower in the 1LOT subcohort as compared with the overall treated cohort, but the cost associated with transplants was highest in the 1LOT subcohort, indicating the application of 1L consolidated therapy among the PTCL population.

Although four US-based studies have previously evaluated HRU or costs in PTCL patients, these economic burden studies either assessed a subset of the PTCL population such as Medicare enrollees, presented findings based on older data (latest data was 2017), or lack comprehensive reporting of HRU categories and associated costs, especially including patients with stem cell transplants [11–14]. Consistent with published studies, hospitalizations and prescription medication utilization were the main cost drivers of all-cause and PTCL-related costs [13,14,16]. Furthermore, the high cost of treatment and disease burden is evident by mean all-cause total healthcare costs PPPM being three-times higher (US$8040 vs $24,924) than corresponding baseline costs, with prescription drugs and hospitalization contributing nearly 70% of the total costs. Compared with a previously reported mean all-cause cost PPPM of US$6327 during the year 2012, the current corresponding cost was nearly four-times higher [14]. Furthermore, PTCL-related HRU and costs accounted for 70% of the total all-cause healthcare costs, demonstrating the significant economic burden associated with PTCL diagnosis and treatment. Within each LOT subcohort, PTCL-related costs accounted for 69.8, 69.3 and 67.2% of total all-cause costs of the 1LOT, 2LOT and ≥3LOT cohorts, respectively. As observed in the sensitivity analysis of total all-cause and PTCL-related healthcare costs, patients with longer follow-up of at least 12 months had lower mean costs compared with the main cohort of patients with at least 1 month of follow-up. This is likely explained by immortal time bias present in the 12-month follow-up requirement, selecting for a healthier population. In comparison to our results for all-cause (US$14,073 PPPM) and PTCL-related costs ($9698 PPPM), a published study by Shah et al. reported lower all-cause ($5885 PPPM) and PTCL-related ($2279 PPPM) costs of patients with at least 12 months of follow-up [13]. However, since the previous study was conducted in Medicare patients with the latest data from 2017, the reported costs are expected to be lower than those estimated in the current study.

This study has limitations inherent to a retrospective study using a claims database, including potential misclassification of diagnosis, data entry errors, lack of clinical data to confirm cancer diagnosis, and reasons for treatment discontinuation or switch. Patients were identified with PTCL using ICD-9 and ICD-10 diagnosis codes, some of which may lack high specificity to PTCL. For example, we observed a lower number of AITL subtypes among PTCL patients than expected. Along with potentially low ICD code specificity, the decreased count may be due to the higher proportion of AITL patients enrolling in clinical trials and lower prevalence of AITL in North America [28]. We also observed a relatively small number of patients in the treated cohort (~20%) compared with the overall cohort, which also may reflect the lack of specificity of ICD codes for PTCL. However, several prior studies have utilized these ICD codes to identify PTCL patients, and our study utilized a more conservative code list by removing several nonspecific ICD-9 codes used in these prior studies [14,16].

The PharMetrics Plus database is representative of a younger, commercially insured population, and thus may not be generalizable to the overall USA, Medicare or Medicaid populations, or uninsured patients. Due to the rare nature of PTCL, only a small sample size of patients was identified who were then further stratified into subgroups based on treatment, thus limiting statistical power. The proportion of treated patients was smaller than previous published studies [13,14,29], but this may be explained by our study definition of systematic treatment which did not include steroid therapy alone. Furthermore, treatments for PTCL were identified a priori, but these treatments may have been utilized for treatment of another disease (e.g., methotrexate has multiple indications for use outside of oncology). Moreover, a claims-based algorithm using gaps in treatment was utilized to classify patients into LOT subcohorts which may not accurately identify specific LOTs. Claims data does not provide reason for treatment gaps and any clinically relevant lapses in treatment may be classified as a distinct LOT. Last, to provide a comprehensive overview of real-world treatment patterns, this study included all patients with ≥1 inpatient or ≥2 outpatient diagnoses (≥30 days apart) for PTCL which could have selected both prevalent and newly diagnosed PTCL patients. For treated patients and their treatment patterns, a 6-month washout period for systemic PTCL treatment was implemented prior to PTCL diagnosis, but patients could have had treatments prior to this period. As treatment regimens differ for newly diagnosed and relapsed/refractory PTCL patients, it may have affected the observed treatment regimens in the current study and may potentially explain the high use of steroid monotherapy.

Conclusion

The findings from this study suggest that treatment patterns observed among PTCL patients included in this study were generally consistent with treatment guidelines, albeit with a high use of monotherapy. High all-cause and PTCL-related HRU and costs were observed in PTCL patients, representing a substantial economic burden associated with PTCL and its management. Given the significant clinical and economic burden of PTCL, novel interventions that improve patient outcomes and reduce the burden of PTCL are needed. Future studies examining the impact of nonsystemic therapies such as stem cell transplant and radiation therapy may provide more insights into the treatment patterns associated with PTCL. The rare nature of PTCL requires additional studies, with larger sample sizes, to fully capture the clinical outcomes associated with PTCL diagnosis and treatment.