Analysis of skin-resident memory T cells following drug hypersensitivity reactions

To the Editor:

Tissue-resident memory T (T_RM) cells persist in peripheral tissues such as skin and do not recirculate (Mueller and Mackay, 2016). T_RM cells protect against infections and mediate anti-tumor responses, but have also been implicated in driving autoimmune and other immune-mediated diseases (Masopust and Soerens, 2019),(Park and Kupper, 2015). However, relatively little is known about their role in mediating delayed drug hypersensitivity reactions. Studies of cutaneous drug reactions such as fixed drug eruption that involve recurrent lesions at the same site after re-exposure to the offending agent, have indicated that localised skin T_RM cells may be mediating inflammatory responses (Shiohara et al., 2002). However, it is not known if T_RM cells play a causative role in other delayed (T cell-mediated) hypersensitivities, such as Stevens-Johnson syndrome (SJS), drug reaction with eosinophilia and systemic symptoms (DRESS), severe maculopapular exanthema (MPE) and other T-cell mediated drug-induced cutaneous diseases. We sought to examine the role of T_RM cells in a prospective cohort of patients with a history of antibiotic-induced DRESS or MPE.

Acute DRESS and MPE are both characterized by a widespread rash involving the skin surface indicating that drug-specific T cell-mediated responses are occurring in the skin. Indeed, we found that a T cell infiltrate of predominantly CD4⁺ T cells in skin sampled from a patient with acute DRESS (Figure S1a). Skin biopsies from a patients presenting with DRESS showed that CD4⁺ and CD8⁺ T cells were present in affected skin after disease onset (acute disease) and following resolution of skin pathology (post-disease), during which T cells persisted at relatively higher frequencies, with the majority of T cells expressing CD69 (Figure S1b, c).

This increase in T cells in the post-disease skin biopsy taken 44 days later suggested that the skin had not yet reached a quiescent state and that persistent smoldering inflammation in the absence of clinically overt disease may be occurring, or may reflect increased cell death in the lesional skin biopsy. To formally assess drug-reactive T cell responses in vivo, we applied the antibiotic implicated in the acute DRESS or MPE reaction intradermally in the recovery phase of the drug reaction and monitored skin for the development of a local response (Trubiano et al., 2019). As previously described (Trubiano et al., 2019), intradermal testing was performed in forearm skin, a site that was involved in the initial drug exanthema. Normal Saline was inoculated intradermally in an adjacent site as a negative control. A positive result was defined as induration of >5mm that persisted for > 72 hours in a patient with a negative Normal Saline control, as per previous definitions (Trubiano et al., 2018). Skin biopsies of the positive intradermal site reaction (‘lesion’) and non-affected control skin (‘control’) were obtained one day (acute IDT) and more than 8 weeks (post-IDT) after intradermal drug inoculation (Figure 1A). Post-IDT biopsies were taken from skin immediately adjacent to the index biopsy in the lesional and control areas. Five patients with a history of DRESS or severe MPE developed a positive intradermal site reaction one day after inoculation with causative drug (Table 1). We performed flow cytometry on lymphocytes extracted from skin biopsies to determine T cell content and expression of CD69 and the integrin CD103, markers that denote tissue-residency in skin (Mackay et al., 2013). Whereas CD69⁺CD103⁻ T cells may represent cells in transit through the tissue (Mackay et al., 2015), CD8⁺CD69⁺CD103⁺ skin T_RM cells are widely held to be a stable non-recirculating population (Mackay et al., 2013, Mackay et al., 2015, Watanabe et al., 2015), although some uncertainty about these distinctions remain (Klicznik et al., 2019). Shortly after drug inoculation, we found increased numbers of CD4⁺ T cells in the inoculated lesion skin site as compared to the control skin (median 4.8-fold increase in lesional skin) although in convalescence, this site-specific localisation of CD4⁺ T cells in lesional skin was reduced (2.8-fold increase in lesional skin) (Figure 1b, c, Figure S2a). In contrast, the number of CD8⁺ T cells did not vary substantially in lesional skin compared to the control skin site during acute and convalescent phases (Figure 1b, c, Figure S2a). Within the CD4⁺ T cell population, cells of a CD69⁺CD103⁻ phenotype dominated lesional skin in the acute phase at a 5.1-fold increase over control skin, although this dominance reduced over time (Figure 1d, 1e, left panel; Figure S2b). Conversely, the ratio of CD4⁺CD69⁺CD103⁺ cells in lesion versus control skin was increased in the convalescence phase (median 2.9-fold increase) indicating conversion to the T_RM cell phenotype over time (Figure 1d, e; Figure S2c). This observation was even more striking for CD8⁺ T cells expressing CD69 and CD103, which were found in lesional skin at a 5.4-fold increase over control skin in convalescence (Figure 1d, e, Figure S2c). Taken together, this data demonstrates that T cells infiltrate and persist as T_RM cells following intradermal drug challenge, supporting the notion that skin T_RM cells contribute to cutaneous drug hypersensitivity reactions.

Figure 1.

T_RM cell persistence in skin following intradermal drug inoculation in patients with a history of DRESS or MPE. IDT with causative drugs was performed on five patients (3 DRESS [circles] and 2 MPE [squares]), and skin biopsies were taken from the reactive exanthema (Lesion) and control (Ctrl) skin 1 day (Acute-IDT) and > 8 weeks (Post-IDT, red) later. Both MPE patients had CLL. (a) Image shows Lesion and Ctrl skin biopsy sites during acute-IDT reaction and post-IDT from a representative patient (ID 1). The site of the control biopsy was distinct from the adjacent lesion. This patient was IDT positive to additional beta-lactam antibiotics (benzylpenicillin, ampicillin), which are routinely tested in the setting of a primary beta-lactam allergy (flucloxacillin). Patient consent for image publication was obtained. (b–e) Flow cytometry was performed on lymphocytes extracted from skin biopsies. Individual data points are shown (n = 5) and paired samples connected where available (n = 3). (b) Representative flow cytometry plots show CD4 and CD8 staining of CD3⁺CD45RO⁺ T cells. The mean frequency is shown (n = 4 Acute-IDT, n = 4 Post-IDT). (c) Ratio of the number of CD4⁺ (left) and CD8⁺ (right) T cells isolated from Lesion compared with Ctrl skin. (d) Representative flow cytometry plots show CD69 and CD103 expression by CD4⁺ (left) and CD8⁺ (right) CD3⁺CD45RO⁺ T cells. The mean frequency is shown (n = 5). (e) Ratio of the number of CD103⁻ or CD103⁺ T cells isolated from Lesion compared with Ctrl skin of CD69⁺ CD4⁺ T cells (left) and CD69+ CD8⁺ T cells (right). CLL, chronic lymphocytic leukemia; Crtl, control; DRESS, drug reaction with eosinophilia and systemic symptoms; IDT, intradermal drug testing; Lesion, reactive exanthema; MPE, maculopapular exanthema; T_RM, tissue-resident memory T cells.

Table 1 –

Clinical features of patients with positive intradermal drug testing

ID	Age/Sex	ICHa	CCI	Pheno-type	RegiSCAR score	Internal organ involved	Primary implicated drugs	Days from phenotype onset to IDT	Positive IDT drugd	Days between Acute IDT and Post-IDT Bx	Time points analyzed by flow cytometry	ELISpot resultb
1	51F	-	1	DRESS	4	No	PTZ, FLU	110	FLUe	105	Acute, conv	Neg
2	70M	CLL	4	MPE	3	No	ADF	57	CLAV	NA	Acute	Neg
3	70F	CLL	5	MPE	1	No	ADF	2118	AMX	63	Acute, conv	Neg
4	41F	-	5	DRESS	7	Yes (Liver, renal)c	PTZ, VAN	1202	PTZ	56	Conv	Pos -PTZ
5	72F	-	3	DRESS	5	Yes (Liver, renal)c	CEF, VANC	861	CEF	63	Acute, conv	Pos- CEF

Abbreviations– M, male; F, female; DRESS, drug reaction with eosinophilia and systemic symptoms; MPE, maculopapular exanthema; CLL, chronic lymphocytic leukemia; CCI, age-adjusted charlson comorbidity index; PTZ, piperacillin-tazobactam; FLU, flucloxacillin; ADF, amoxicillin clavulanate; CLAV, clavulanate; AMX, amoxicillin; VAN, vancomycin; CLI, clindamycin; CEF, ceftriaxone; NP, not performed; NA, not applicable; Neg, negative; Pos, positive; Conv, convalescent; Bx, biopsy; IDT, intradermal testing; NA, not applicable

^aICH, immunocompromised host – HIV, cancer, haematological malignancy, transplant, autoimmune/connective tissue disorder, pred > 10mg daily for 1 month

^bIFN-γ release ELISpot performed to implicated drugs from PBMCs (see Supplementary Materials and Methods)

^cLiver injury defined as per drug induced livery injury definition (DILI) (Aithal et al., 2011) and renal injury as per a 50% increase in creatine from baseline or requirement of continuous renal replacement therapy.

Following intradermal drug inoculation, the acute exanthema develops rapidly. This response could be mediated by local or transiting skin cells that respond to cognate antigen, or bystander T cells that may have accumulated in the previously affected site. To assess whether circulating drug-reactive T cells were contributing to development of acute exanthema after intradermal testing, we measured circulating T cell responses to the implicated drug ex vivo using IFN-γ release ELISpot (Trubiano et al., 2018) at the time the intradermal testing (IDT) was administered or in convalescence. Only two patients had reactive IFN-γ release ELISpot tests, both of which had severe multiple internal organ involvement at presentation (Figure S3, Table 1). However, both uniquely had severe multiple internal organ involvement at presentation, suggesting that circulating T cells may be related to organ involvement rather than contributing to the lesional skin response.

Severe T cell mediated hypersensitivity reactions are associated with significant morbidity and mortality (Trubiano et al., 2016) yet the immunopathogenesis remains ill-defined and there is no consensus regarding optimal therapy. Previous case reports have noted an absence of circulating drug-reactive T cells in severe T cell mediated cutaneous reactions (Iriki et al., 2014) raising the possibility that skin-resident T_RM cells are key mediators of such skin hypersensitivity reactions. Moreover, recent work has shown a reduction in T_RM cell function after immune checkpoint blockade therapy may reduce the severity of the non-drug hypersensitivity disease allergic contact dermatitis (Gamradt et al., 2019). Here, our findings show that skin T cells persist and covert to the T_RM cell phenotype following intradermal drug challenge, supporting the hypothesis that skin T_RM cells may contribute to cutaneous drug hypersensitivity. This provides previously unreported insights into the pathogenesis of T cell-mediated cutaneous hypersensitivity reactions.

Abbreviations:

M

male

F

female

DRESS

drug reaction with eosinophilia and systemic symptoms

MPE

maculopapular exanthema

CLL

chronic lymphocytic leukemia

CCI

age-adjusted charlson comorbidity index

PTZ

piperacillin-tazobactam

FLU

flucloxacillin

ADF

amoxicillin clavulanate

CLAV

clavulanate

AMX

amoxicillin

VAN

vancomycin

CLI

clindamycin

CEF

ceftriaxone

NP

not performed

NA

not applicable

Neg

negative

Pos

positive

Conv

convalescent

Bx

biopsy

IFN-γ

interferon gamma

ICH

immunocompromised host

TRM

tissue-resident memory T cells