MEASURING THERAPEUTIC RESPONSE IN CHRONIC GRAFT-VERSUS-HOST DISEASE

Skin and skin appendages

Skin is the most frequently affected organ in chronic GVHD, and manifestations are highly variable. In the 2005 response criteria, proposed measures included the percentage of body surface area (BSA) by the type of involvement (erythematous rash, movable sclerosis, non-moveable sclerosis). BSA measurement on a continuous scale suffered from poor inter-rater reliability, particularly for moveable sclerosis.³² Thus, the 2014 revision recommends the simpler NIH Skin Score instead, because it correlates with chronic GVHD severity, symptoms, and survival.³⁴ The NIH Skin Score is a 0–3 score that summarizes BSA involvement into four categories: no skin involvement; ≤18% without sclerosis; 19–50% or any moveable sclerosis; and >50% or any non-movable sclerosis, impaired mobility, or ulcers. The “Rule of 9’s” as an estimate of BSA involvement is intended for use in adults and is less accurate in children, particularly young children. For the sake of simplicity, we recommend using the “Rule of 9’s” for all children, except for those less than one year of age. A body surface area grid for children less than 1 year of age can be found at http://www.asbmt.org/?page=MeasureCGVHD. BSA assessment should include superficial skin eruptions, moveable sclerosis and non-movable sclerosis. Superficial skin eruptions of cutaneous chronic GVHD include maculopapular, erythematous, lichen planus-like, papulosquamous, ichthyotic, and keratosis pilaris-like rashes. Superficial sclerosis (moveable) includes both lichen sclerosis-like and morphea-like lesions. Deep sclerosis (non-movable) includes diffuse, fixed (hidebound) sclerosis, fibrosis of subcutaneous fat septae (“rippling”) and fasciitis (“groove sign”). The presence of ulcers should be noted but documentation of the size is no longer required.

Sclerotic changes are common in skin GVHD, difficult to measure reliably, and respond slowly to therapy. Since quantitative methods to measure the depth of sclerotic involvement are not available in a general oncology practice, these changes have been described in more qualitative terms related to thickening, pliability, color, adherence to underlying tissues, or changes in joint mobility. No validated scale exists for assessing sclerotic skin changes of chronic GVHD. Measures such as the Rodnan scale for assessment of systemic sclerosis might be helpful for clinical evaluation, but this scale is not suitable for use in clinical trials because it does not address the full spectrum of sclerotic skin manifestations in chronic GVHD. There is an urgent need for the development of more quantifiable and reproducible measurements or imaging methods that could be used in patients with sclerotic skin manifestations of chronic GVHD.^14–18 Pending development of these measures, the Working Group recommends recording the percentage of BSA involved with deep sclerosis/fasciitis and using an exploratory 0–10 semi-quantitative scale for capturing clinician-perceived severity of sclerosis using the descriptor “skin and/or joint tightening.”

Pigmentary changes do not indicate chronic GVHD disease activity per se. Moreover, changes in pigmentation occur gradually and are perceptible only across long time intervals. Thus, these changes are not scored for the purposes of response assessment.

Patient-reported measures of skin disease are also useful for measuring response. The skin subscale of the Lee Symptom Scale correlates with severity of skin disease, and changes in patient-reported skin symptoms correlate with survival.³⁴ Patients should report their most severe itching during the past week, rated according to a 0 – 10 scale, since itching is the most frequent cutaneous symptom of chronic GVHD. These are considered recommended measures. A semi-quantitative exploratory measure, “Your skin and/or joint tightening at its worst,” on a 0–10 scale has been added to capture severity of skin sclerosis. Informal cognitive testing suggested that including both “skin” and “joint” tightening in the question was not confusing for patients.

Musculoskeletal connective tissue

Assessment of joint range of motion is a very useful objective measure of chronic GVHD tissue response in patients with sclerotic changes involving large joints. The NIH Joint Score and the photographic range of motion (P-ROM) scales correlate with change in joint involvement, and are recommended measures.³⁵

Eyes

Dry eyes represent either lacrimal gland dysfunction or destruction that may be permanent. Although the Schirmer’s test³⁶ was recommended in the 2005 criteria, subsequent studies have not supported the validity of this test in ocular chronic GVHD monitoring, and it is no longer recommended as a response measure for general chronic GVHD clinical trials.³⁷ The NIH Eye Score is a recommended measure, since it detects improvement or worsening in ocular chronic GVHD.³⁷ Scoring from 0–3 is based on symptoms, need for eye drops, and use of therapeutic procedures or devices.

Patients should report a “chief eye complaint” rated according to a 0 –10 scale for peak severity during the past week. The complaint can change from visit to visit, but only one “chief eye complaint” is graded. This method is simple to use but may impose undesirable limitations in patients with multiple complaints. The eye subscale of the Lee Symptom Scale and the Ocular Surface Disease Index are also sensitive to change,³⁷ but the Lee eye subscale is more convenient since it is shorter and is already included in the composite symptom battery.

Mouth

Previously, oral chronic GVHD was assessed using a NIH modification of the Schubert Oral Mucosa Rating Scale²⁵ (OMRS) that scores oral surfaces from 0 – 15, with higher scores indicating more severe involvement. The four 2005 NIH Consensus chronic GVHD manifestations assessed in this scale included a) mucosal erythema (color intensity and percent of oral surface area); b) lichen-like hyperkeratosis (percent of oral surface area); c) ulcerations (percent of oral surface area); and d) mucoceles (total number). Subsequent studies have suggested that mucoceles are not reliably assessed,^32,33,38 and their enumeration does not correlate with important clinical outcomes.^38,39 Thus, the Working Group recommends removing mucoceles from the NIH-modified OMRS, resulting in a further modified 0–12 scale. The term “hyperkeratosis” has been removed and clarified as “lichen-like changes” instead.

Patients should report their mouth sensitivity (irritation resulting from normally tolerated spices, foods, liquids, or flavors), rated according to a 0 – 10 scale for peak severity during the past week. Children may have an easier time with a 0–3 scale, but this format has not been validated and may be less sensitive to change. Evaluation of mouth dryness and mouth pain on 0–10 scales is no longer recommended for general chronic GVHD trials.⁴⁰ Mouth symptoms are also captured in the Lee Symptom Scale, as part of a recommended battery.

Hematopoietic

Parameters to be documented at trial enrollment include platelet count⁴¹ and absolute eosinophil count,^42,43 since they may have prognostic significance. These measures are not part of the response assessment, and their ongoing documentation is no longer recommended.

GI tract

Gastrointestinal (GI) symptoms are difficult to measure quantitatively. GI symptoms during the preceding week are graded through interview by the examining clinician according to 0 – 3 severity scales for the upper and lower GI tract and esophagus. Patients with chronic GVHD often have weight loss that is not explained by GI symptoms.⁴⁴ Although the exact relationship between weight loss and chronic GVHD activity is not clear, recording patient weight at each scheduled evaluation is strongly encouraged, given the simplicity of this measure and its potential importance for monitoring the success of therapy.

Liver

Liver injury should be assessed according to the most recent laboratory results for total serum bilirubin (mg/dL), alkaline phosphatase (U/L) and alanine aminotransferase (U/L). Laboratory upper limits of normal should also be recorded. Aspartate aminotransferase is not recorded since it is not specific for liver inflammation.

Lung

The 2005 response criteria recommended the lung function score (LFS), based on percent predicted forced expiratory volume in the first second (%FEV1) and single breath diffusion lung capacity for carbon monoxide (DLCO) adjusted for hemoglobin, because it is predictive of respiratory failure and mortality after allogeneic hematopoietic stem cell transplantation.^45,46 Experience in patients with steroid-refractory GVHD has suggested that the LFS is sensitive to change and is useful as a response measure.⁴⁷ On the other hand, DLCO is not directly affected by bronchiolitis obliterans syndrome (BOS), and pulmonary function tests (PFTs) did not perform as well as the NIH Lung Symptom Score in predicting non-relapse mortality and survival in an observational study, although conclusions were limited by 50% missing PFT data.⁴⁸ PFTs cannot be performed in children younger than 5 years of age, and DLCO usually cannot be measured in children younger than 10 years. FEV1 may not be reliable in children.⁴⁹ Another therapeutic study that included mostly BOS patients showed that a decrease in %FEV1 or FVC of greater than 10% was highly correlated with an increased risk of death within 5-years.⁵⁰

The Working Group recommends recording the %FEV1, FEV1 (L), and forced vital capacity (FVC), and strongly encourages parallel documentation of DLCO corrected for hemoglobin, total lung capacity (TLC), and residual volume (RV) to allow further validation studies, including exploration of restrictive lung disease as a manifestation of chronic GVHD.^51–53 If available, the %FEV1 value should be prioritized for response assessment. However, if PFTs are not available, the NIH 0–3 Lung Symptom Score may be used for response assessment. Of note, airway hyper-responsiveness in, and variable cooperation by, pediatric patients may require repeated PFTs in cases where abnormal values are observed. %FEV1 may decline due to respiratory infections or progressive restrictive physiology due to extrinsic thoracic compression from sclerosis or respiratory muscle weakness, and these causes should be considered if PFTs worsen. Exploration of the %FEV1 slope as an outcome measure to account for the change in disease trajectory is encouraged.^54,55 Chronic GVHD of the lung is not a bronchodilator-responsive process, so bronchodilators are not required to assess response for general chronic GVHD trials. Patients with reactive airways disease such as asthma, who have a documented bronchodilator response, should continue to receive bronchodilators during PFTs.

Genitals

Women should be asked specific questions relating to vulvar and vaginal symptoms, such as burning, pain, discomfort, or dyspareunia. Patients who report problems should be referred to a gynecologist experienced in the care of patients with chronic GVHD. Since such symptoms could be under-reported or caused by conditions other than chronic GVHD,⁵⁶ and because proper evaluation requires a specialist exam, measures of genital response are considered exploratory. Both female and male genital symptoms may be captured by the exploratory item rating “Worst genital discomfort” on a scale from 0–10 in adolescent and adult patients while this item will not work in pediatric patients. Academic gynecologists interested in chronic GVHD are developing precise vulvo-vaginal assessment scales. These scales will be useful in selected trials where vulvar and vaginal changes are the primary endpoints of interest.^26,27

Other organ systems may be affected by chronic GVHD, but are either uncommon or difficult to quantify by non-specialists. The Working Group encourages investigators to develop and validate response assessment tools that could detect meaningful clinical benefit in trials focused on specific organs or manifestations.

Clinician perceptions

Physicians, nurse practitioners, or physician assistants should provide a subjective assessment of current overall chronic GVHD severity on a 4-point scale (no chronic GVHD, mild, moderate, severe)⁷ without knowledge of the calculated NIH global severity score. They should also provide an assessment of current overall chronic GVHD severity on an 11-point numerical scale (0 indicates no GVHD manifestations; 10 indicates most severe chronic GVHD symptoms possible).⁶² The categories of mild, moderate, and severe have been used in previous studies for patient and clinician assessment, where they were often undefined but showed good prognostic ability.^7,63 Clinicians should also provide their evaluations of chronic GVHD changes since the last assessment scored on a 7-point scale (very much better, moderately better, a little better, about the same, a little worse, moderately worse, very much worse).⁹ These semi-quantitative assessments may detect qualitative improvements that are clinically meaningful but not well captured using other measures.⁶⁴ The protocol and case report form should specify the time interval for assessing changes.

Patient perceptions

Similarly, at each assessment, patients should score their perceptions of overall chronic GVHD severity, overall severity of symptoms, and change in symptom severity compared with the last assessment, using the same response options used by clinicians.

The exact role of global scales in chronic GVHD response assessments and their appropriate use as outcome measures in clinical trials remain to be determined. These scales could be sensitive to qualitative changes that might otherwise escape detection if the assessments were limited to quantitative measures. They are used in studies that establish clinically meaningful differences in measures. A potential limitation is that personality traits can influence patient perceptions or self-report.⁶⁵

Human Activity Profile

The strongly encouraged patient-reported measure of physical activity in adults is the Human Activity Profile (HAP) questionnaire. The 94 questions are presented in ascending order according to the metabolic equivalents of oxygen consumption required to perform each activity.¹³ The HAP therefore provides a survey of activities that the patient performs independently across a wide range of metabolic demand, beginning with getting out of bed, bathing, dressing, performing a series of progressively more physically demanding household chores, and ending with running or jogging 3 miles in 30 minutes or less. While the HAP correlates with chronic GVHD severity,^72,81 it may not be necessary if the SF-36 is collected, since the SF-36 also assesses physical activity. The Working Group no longer recommends the Activities Scale for Kids (ASK)^82–84 due to lack of data in patients with chronic GVHD.

Performance scales

The Karnofsky or Lansky Performance Scale is commonly used in clinical assessments of chronic GVHD and has prognostic value for survival, so it is strongly encouraged at enrollment.^85,86 These scales are not valid measures of response.

Overall response

Three general categories of overall response are proposed for interpretation of clinical trials: complete response, partial response, and lack of response (unchanged, mixed response, progression). Complete response is defined as resolution of all manifestations in each organ or site, and partial response is defined as improvement in at least one organ or site without progression in any other organ or site as described in the following sections on organ response. The Working Group recommends that skin, mouth, liver, upper and lower GI, esophagus, lung, eye, and joint/fascia be considered in evaluating overall response. Genital tract and other manifestations are not included due to lack of validated response measures.

Complete organ response

The term “complete organ response” indicates resolution of all manifestations related to chronic GVHD in a specific organ. This category may not apply to organs with irreversible damage. The Working Group extensively discussed whether a “near CR” category was justified to indicate resolution of all reversible organ manifestations despite residual minor abnormalities. However, no consensus could be reached on the definition of “irreversibility,” and there is no evidence yet that the ultimate outcome after CR, near CR or PR differs. For these reasons, no “near CR” category is included, and both CR and PR are considered to be meaningful responses in clinical trials.

Partial organ response

The proposed definition of partial response in a specific organ requires an improvement in score from baseline that reflects genuine clinical benefit and exceeds the measurement error of the assessment tool: an improvement of 1 or more points on a 4–7 point scale, or an improvement of 2 or more points on a 10–12 point scale. Partial response in the liver requires at least a 50% improvement in ALT, alkaline phosphatase or total bilirubin. For patients with BOS, an absolute improvement in %FEV1 of 10% predicted or more (e.g., 50% to 60%) is considered a partial response⁵⁰ as long as the initial %FEV1 is <70%. Normalization (>80%) is considered a complete response.

Because the NIH 0–3 Skin Score is now recommended for response assessment, even substantial improvement in sclerotic features will not be considered responses unless the NIH Skin Score improves. To aid in future measure development and validation, recording the body surface area involved with non-moveable sclerosis and collecting the severity of sclerosis on a semi-quantitative 0–10 scale is strongly encouraged. Trials targeting sclerotic chronic GVHD should use more detailed measures to document change in extent and functional consequences of sclerosis.

Organ Progression

Criteria for progression in each organ must be defined, since the overall category of partial response requires the absence of progression in any organ. For skin, eye, esophagus and upper and lower GI tract, a worsening of 1 point or more on the 0–3 scale is considered progression, except a change from 0 to 1, which is considered trivial progression since it often reflects mild, non-specific, intermittent, self-limited symptoms and signs that do not warrant a change of therapy. One study showed poor test-retest reliability in 0 to 1 GI changes, suggesting these minimal changes are also not reliably measured.⁹⁷ For joint/fascia, a worsening of 1 point or more on the 0–3 scale is considered progression, even if from 0 to 1, because a change to score 1 was considered meaningful progression that would prompt a change in therapy. For joints assessed by the P-ROM, a worsening of 1 or more points for the 7 point scales and 1 or more points for the 4 point scale is considered progression. For mouth, a worsening of 2 or more points on the 12 point scale indicates progression. Worsening of liver GVHD is defined by an increase of 2 or more times the ULN for the assay for ALT, alkaline phosphatase or total bilirubin. For patients with lung involvement, absolute worsening of FEV1 by 10% predicted or more (e.g., 50% to 40%) is considered progression, as long as the final %FEV1 is <65% (since initial %FEV1 must be <75% to establish the diagnosis of lung involvement). If PFTs are not available, then worsening of the clinical lung score based on symptoms by 1 or more points should be scored as progression, except from score 0 to 1, which is considered trivial progression due to its lack of specificity for lung chronic GVHD. If the clinical lung symptom score worsens from 0 to 1, PFTs are encouraged if lung chronic GVHD is suspected.

For all organs, progression cannot be scored for manifestations with baseline values that are too close to the worst score, especially the organs using 0–3 scales, raising the possibility that patients could experience significant clinical worsening in an organ and still be considered a PR if they have measurable improvement in another organ. However, this concern is mitigated because many clinicians would change therapy in such a situation, and any addition of a new systemic immunosuppressive treatment is considered as indicating a lack of response.

Mixed Response

Mixed response is a new category defined as complete or partial response in at least one organ accompanied by progression in another organ. This category should be considered progression for the purposes of analysis but may aid in identifying organ-specific response patterns.

Unchanged

Outcomes that do not meet the criteria for complete response, partial response, progression or mixed response are considered unchanged. These outcomes will generally be considered non-responders unless specified otherwise by the protocol.

Limitations in measurement of organ responses

The response criteria do not account for qualitative changes. Clinical experience indicates that clinically important qualitative improvement often occurs before improvement in the objective measures. For this reason, the response criteria are not intended for use as the primary guide for clinical decisions. Certain organs or rare manifestations are not considered in the response criteria because quantitative assessments are not feasible, but these may be the most important manifestations of chronic GVHD for individual patients. To capture qualitative and global changes of the entire chronic GVHD syndrome, use of the clinician and patient assessed 0–10 global scale is strongly encouraged. The response criteria also do not account for the prior trajectory of abnormalities. For example, “stable” or “unchanged” disease might be considered a meaningful response when the prior trajectory was clear progression, as indicated, for example, by serial pulmonary function tests or rapidly progressive sclerosis, whereas “stable disease” after prior improvement or stability should not be considered a “response.” If the prior trajectory is important for response determination, the protocol should specify data to be documented in order to establish the trajectory.

Validation of response criteria

One study showed significant prognostic value of the NIH calculated responses for predicting survival in the context of a therapeutic trial⁴⁷ but this conclusion was not supported by results from a large observational study.⁹⁸ The criteria proposed in these guidelines have been modified based on publications since 2005 but still need to be validated prospectively for patients with chronic GVHD. For these reasons, the updated criteria are still provisional and subject to change with further clinical experience. Also, depending on the stringency of response definitions required by the specific study, these general guidelines could be modified to fit the needs of a particular protocol. Since the criteria are subject to change, we strongly recommend that case report forms should always record the actual numerical values for any measurement in order to allow future recategorization.

Caveats for assessment of response in clinical trials

Protocols must specify the times when response will be assessed, generally 2–6 months after enrollment, and the requirement(s) for durability of response. Permanent discontinuation of systemic chronic GVHD therapy confirmed for the duration of the observation period indicates a durable response. In principle, if additional systemic therapy for chronic GVHD is added before the end of the specified study period, the outcome is categorized as lack of response. The protocol should specify whether dose changes, for example increased steroid doses, are considered “additional systemic therapy.” Specific reasons for additional systemic therapy should be collected, since new treatments are sometimes added even though chronic GVHD is stable or improving (e.g., loss or change of insurance coverage, transportation or logistical issues, loss of central IV access etc.), and protocols should specify whether these cases should be categorized as lack of response.

Occasionally patients have co-morbid conditions that may interfere with response assessment (for example, a viral infection at the time of a planned PFT, temporary cholestasis due to a gallstone) or pre-existing irreversible conditions (for example post traumatic joint contracture, lung disease etc). In these situations the abnormality should still be scored but the check box (From A) should also indicate that the abnormality is explained entirely by a non-GVHD documented cause and the cause should be specified. Protocols should specify whether response measures may be delayed to allow resolution of a non-GVHD issue before response assessment or whether an organ entirely explained by another non-GVHD cause should be excluded for organ-specific or overall response assessment.

Addition of topical or organ-directed treatments generally make it impossible to be sure any response is due to systemic immunosuppressive treatment. Thus, topical and local therapy added after study enrollment must be documented in all patients and considered especially in patients who have a complete or partial response. Sensitivity analyses may be appropriate to determine whether exclusion of responses in the affected organs or sites changes the study conclusions. Some topical or organ-directed therapies may result in systemic effects due to absorption (e.g., oral, GI tract, lung, skin) or effects beyond the target organ (e.g., montelukast), and protocols should specify whether such treatments are allowed during the course of the trial or would automatically indicate a lack of response.