Karp et al. (2012): Randomized, double-blind, placebo-controlled, phase III chemoprevention trial of selenium supplementation in patients with resected stage I non-small-cell lung cancer: ECOG 5597

Reference ↗
Title	Randomized, double-blind, placebo-controlled, phase III chemoprevention trial of selenium supplementation in patients with resected stage I non-small-cell lung cancer: ECOG 5597
Topic	Selenium
Author	Karp, DD, Lee, SJ, Keller, SM, Wright, GS, Aisner, S, Belinsky, SA, Johnson, DH, Johnston, MR, Goodman, G, Clamon, G, Okawara, G, Marks, R, Frechette, E, McCaskill-Stevens, W, Lippman, SM, Ruckdeschel, J, Khuri, FR
Year	2012
Journal	Journal of clinical oncology: official journal of the American Society of Clinical Oncology
DOI	https://doi.org/10.1200/JCO.2013.49.2173

Brief summary

In this study, 1561 participants with resected non-small cell lung cancer were included. They were randomly divided into two arms and were each to take 200 µg of yeast selenium (1040 participants) or a placebo (521 participants) every day. However, the study was terminated early, as an interim analysis after a study period of around nine years showed that no benefit could be achieved for the selenium arm; instead, there was a positive trend for the placebo arm. In a final analysis another two years later, there were no significant differences in the incidence of secondary tumors, lung cancer or other cancer recurrence, time to recurrence or overall survival of the patients. However, it was shown that participants who had never smoked had a higher survival time after three and five years than those who were current smokers or had only quit in the last year. The study leaves out basic framework descriptions. A promising subgroup analysis between different selenium concentrations and their effects on the cancer-free time interval remains without statistical value and thus cannot be clearly interpreted.

In dieser Studie wurden 1561 Probanden mit reseziertem nicht-kleinzelligem Lungenkarzinom eingeschlossen. Diese wurden zufällig in zwei Gruppen eingeteilt und sollten jeweils jeden Tag 200 µg Hefeselen (1040 Probanden) oder ein Placebo (521 Probanden) zu sich nehmen. Die Studie wurde allerdings frühzeitig abgebrochen, da sich in einer Zwischenanalyse nach etwa neun Jahren Studienlaufzeit abzeichnete, dass kein Vorteil für die Selengruppe erzielt werden könne, stattdessen gab es einen positiven Trend für die Placebogruppe. In einer abschließenden Analyse weitere zwei Jahre später zeigten sich keine bedeutsamen Unterschiede bezüglich des Auftretens von sekundären Tumoren, Lungenkrebs oder sonstiger Rückkehr von Krebs, sowie der Zeit bis zur Rückkehr oder dem Gesamtüberleben der Patienten. Allerdings wurde gezeigt, dass Personen, die nie geraucht haben, eine höhere Überlebenszeit nach drei und fünf Jahren zeigten, als Probanden die aktuell rauchten, oder erst im letzten Jahr aufgehört hatten. Die Studie vernachlässigt grundlegende Rahmenbeschreibungen. Eine vielversprechende Subgruppenanalyse zwischen verschieden hohen Selenkonzentration und deren Auswirkungen auf das Krebsfreie-Zeitintervall bleibt ohne statistische Wertangabe und kann damit nicht eindeutig interpretiert werden.

Study Design

Prospective / Retrospective Prospective: forward-looking, examples include clinical trials, cohort studies, and long-term observational studies;</br>Retrospective: backward-looking, relying on existing data, examples include case-control studies and retrospective cohort studies	Prospective
Monocentric / Multicentric Monocentric: conducted in one center/ hospital; </br>Multicentric: conducted in multiple centers/ hospitals	NI
Blinding No: Open, all parties are aware of group assignments;</br>Single: one party is unaware of group assignments (generally participants);</br>Double: two parties are unaware of group assignments (generally the participants and the researchers); </br>Triple: concealing group assignment from additional parties	Double
Is randomized	Yes
Cross-over Participants alternate between different treatment groups or conditions over a specified period, allowing each participant to serve as their own control	No
Number of arms	2

Study characteristics

Inclusion criteria	36 months from complete resection of histologically proven stage IA (pT1N0) or stage IB (pT2N0) NSCLC (carcinoid tumors were excluded); pathologic stage N0 confirmed by sampling at least one mediastinal lymph node at resection; chest x-ray or computed tomography scan ≤ 8 weeks before registration without sign of new or recurrent lung cancer; no concurrent cancers or any other prior cancer history within the past 5 years, except localized nonmelanoma skin cancer; no synchronous lesions (lung + nonlung) or metastasis, even if resectable; no history of greater than one lung cancer primary tumor at any time; normal hepatic function (total bilirubin and Aspartat-Aminotransferase or Alanin-Aminotransferase ≤ institutional upper limit of normal); laboratory values (incl. Complete Blood Count) obtained within 8 weeks before registration; and Eastern Cooperative Oncology Group performance status of 0 or 1. Patients with stage IA Non-Small Cell Lung Cancer should not have received any therapy other than surgery. Patients with stage IB Non-Small Cell Lung Cancer were allowed to have received other primary therapy (chemotherapy, radiotherapy, or biologic therapy) provided this was completed at least 6 months before study registration and all treatment-related symptoms had subsided before study registration. Supplements were defined as any nonfood compound taken by mouth or injection to provide dietary factors. Supplements containing ≥ 70 µg of selenium taken regularly (≥ three times per week for ≥ 4 consecutive weeks during the prior year) were required to be discontinued ≥ 1 month before registration. Supplements containing ≤ 70 µg of selenium were continued throughout study participation. Supplements not containing selenium were either discontinued ≥ 2 weeks before study entry or continued throughout study participation.
Exclusion criteria	NI
N randomized	1561
Analysis PP: Per Protocol analysis, i.e. only participants included who adhered to the study protocol.</br>ITT: Intention-to-treat analysis, i.e. all randomized participants included regardless of any drop-outs or changes in assignment.</br>mITT: modified Intention-to-treat analysis can refer to analyses in which participants with missing outcome data are excluded or it can refer to analyses in which only participants who received at least one treatment dose are included. In this case, participants dropped out of the study prematurely for reasons unrelated to the treatment.	ITT Analysis
Specifications on analyses	NA
Countries of data collection	United States
LoE Level of evidence	2b Oxford 2009
Outcome timeline Data collection times	Study period: October 6, 2000 - November 5, 2009

Characteristics of participants

Setting Refers to cancer therapy setting.</br>- Curative therapy: aims to completely eradicate a disease and achieve a full recovery; </br>- Neo-adjuvant therapy: form of curative therapy, given before the primary treatment for cancer (usually surgery); </br>- Adjuvant therapy: form of curative therapy, given after the primary treatment for cancer (usually surgery); </br>- Palliative therapy: focuses on providing relief from symptoms and improving the quality of life for patients, without necessarily targeting the underlying disease; </br>- Active surveillance: involves close monitoring of disease progression without any intervention (typically used for prostate cancer);</br>- No therapy setting: Patients who completed therapy/are currently not in cancer treatment, cancer survivors.	No therapy setting
Types of cancer "Other Cancers" means that only a subpopulation was specified, but further unspecified cancer types were included	Lung Cancer - Non-Small Cell Lung Cancer
Cancer stages Early Stage: generally refers to cancer that is localized to the area where it started, mostly stages I and II;</br>Advanced Stage: cancer that has spread beyond its original site, mostly stages III and IV, with stage IV indicating distant metastasis	Early Stage
Specifications on cancer stages	IA (pT1N0) or IB (pT2N0)
Comorbidities	NI
Current cancer therapies	No therapy
Specifications on cancer therapies	NA
Previous cancer therapies	Surgery, Chemotherapy, Radiation therapy
Gender	Mixed
Gender specifications	Female: intervention arm 51%, placebo arm 52%
Age groups	Adults (18+)
Age groups specification	Age (median): 66 years; range: intervention arm 24-94 years; placebo arm 38-86 years

Arms

Arm type Active control: group receives active treatment; </br>Passive control: for example treatment as usual, waiting control, no treatment	Intervention
Number of participants (arm) N randomized	1,040
Drop-out Number of participants who left the study for any reason or did not provide information on every data collection date	N=817
Drop-out reasons	Ineligible (n = 10); No therapy (n = 12) Discontinued treatment: Completed (n = 238); Progression (n = 117); Toxicity (n = 20); Death (n = 18); Withdrawal (n = 157); Nonprotocol treatment (n = 1); Complicating disease (n = 30); Maximum dose (n = 2); Other (n = 212)
Intervention	Selenium
Dosage and regime	200 µg daily
One-time application	No
Duration in days For long-term interventions, the number of days is an estimation.</br>A value of -999 indicates that the exact duration cannot be extracted from the study due to ambiguous or incomplete information. See Outcome timeline or Dosage and regime for further information.	1,460
Side effects / Interactions	Unclear association, but in long-term follow-up: n=26 in intervention arm had a diabetes diagnosis

Arm type Active control: group receives active treatment; </br>Passive control: for example treatment as usual, waiting control, no treatment	Placebo
Number of participants (arm) N randomized	521
Drop-out Number of participants who left the study for any reason or did not provide information on every data collection date	N=406
Drop-out reasons	Ineligible (n = 6); No therapy (n = 2) Discontinued treatment: Completed (n = 134); Progression (n = 52); Toxicity (n = 8); Death (n = 12); Withdrawal (n = 67); Nonprotocol treatment (n = 3); Complicating disease (n = 17); Other (n = 105)
Intervention	Placebo
Dosage and regime	Daily
One-time application	No
Duration in days For long-term interventions, the number of days is an estimation.</br>A value of -999 indicates that the exact duration cannot be extracted from the study due to ambiguous or incomplete information. See Outcome timeline or Dosage and regime for further information.	1,460
Side effects / Interactions	Unclear association, but in long-term follow-up: n=12 in placebo arm had a diabetes diagnosis

Outcomes

DFS (Disease-Free Survival)

Outcome type As specificed by the authors	Primary
Outcome specification	DFS Randomization until secondary tumors or recurrence and 5-year DFS
Type of measurement	Observation
Results during intervention - Results during intervention means that the time of data collection is during or shortly after the period of the intervention (e.g. on the last day or a few days after). The results therefore still relate to the direct effects of the intervention.</br>- Results after intervention means there is a longer break between the time of data collection and the end of the intervention, e.g. more than a week. The results relate more to long-term effects.</br>- If a categorization in Results during vs. after intervention is not possible (e.g. survival data), the results are summarized under Results after intervention under the headline "Overall".	During study: n=44 non-melanoma skin cancers, (n=14 vs. n=13 basal cell carcinoma; n=11 vs. n=6 squamous cell carcinoma in intervention vs. placebo arm), no significance values given
Results after intervention - Results during intervention means that the time of data collection is during or shortly after the period of the intervention (e.g. on the last day or a few days after). The results therefore still relate to the direct effects of the intervention.</br>- Results after intervention means there is a longer break between the time of data collection and the end of the intervention, e.g. more than a week. The results relate more to long-term effects.</br>- If a categorization in Results during vs. after intervention is not possible (e.g. survival data), the results are summarized under Results after intervention under the headline "Overall".	Interim analysis October 2009: N=83 secondary lung tumors; Rate intervention arm: 1.91 per 100 persons per year vs. 1.36 in placebo arm; p=ns June 2011: N=252 secondary tumors in 224 patients, of which 98 (out of 97 patients) were lung cancer (38.9%); Lung cancer and other cancers (SPT) in intervention arm: 1.62 and 3.54 per 100 persons per year vs. placebo arm: 1.30 and 3.39 per 100 persons per year; p=0.294 5-year DFS: October 2009: Intervention arm: 72% vs. placebo arm: 78%, no significance values given June 2011: Intervention arm: 74.4% vs. placebo arm: 79.6%; p=0.69 Note: 5-year DFS rates with subdivision into low, average and high selenium concentration intervention vs. placebo: 75.5% (SE, 10.3%) vs. 72.9% (SE, 12.7%), 75.6% (SE, 2.27%) vs. 78.2% (SE, 3.3%), and 72.9% (SE, 4.5%) vs. 80.9% (SE, 5.2%)

Risk of Bias Assessment: Cochrane RoB tool 2.0
Bias arising from the randomization process	low risk
Bias due to deviation from intended intervention (assignment to intervention)	low risk
Bias due to deviation from intended intervention (adhering to intervention)	NA
Bias due to missing outcome data	low risk
Bias in measurement of the outcome	low risk
Bias in selection of the reported result	low risk
Other sources of bias	some concerns
Overall RoB judgment	low risk

OS (Overall Survival)

Outcome type As specificed by the authors	Primary
Outcome specification	5-year OS
Type of measurement	Observation
Results during intervention - Results during intervention means that the time of data collection is during or shortly after the period of the intervention (e.g. on the last day or a few days after). The results therefore still relate to the direct effects of the intervention.</br>- Results after intervention means there is a longer break between the time of data collection and the end of the intervention, e.g. more than a week. The results relate more to long-term effects.</br>- If a categorization in Results during vs. after intervention is not possible (e.g. survival data), the results are summarized under Results after intervention under the headline "Overall".	NA
Results after intervention - Results during intervention means that the time of data collection is during or shortly after the period of the intervention (e.g. on the last day or a few days after). The results therefore still relate to the direct effects of the intervention.</br>- Results after intervention means there is a longer break between the time of data collection and the end of the intervention, e.g. more than a week. The results relate more to long-term effects.</br>- If a categorization in Results during vs. after intervention is not possible (e.g. survival data), the results are summarized under Results after intervention under the headline "Overall".	Intervention arm: 76.8% (SE, 1.6%) vs. placebo arm: 79.9% (SE, 2.1%); p=0.154 Note: Distribution significantly different for smoking status (p=0.027): active smokers or those who had quit within one year showed a 3-year OS of 85.5% (SE, 1.7%) and 5-year OS of 74.9% (SE, 2.4%), whereas participants who had never smoked showed a 3-year OS of 90% (SE, 2.8%) and a 5-year OS of 83.6% (SE, 3.6%); no difference for DFS; p=0.245

Risk of Bias Assessment: Cochrane RoB tool 2.0
Bias arising from the randomization process	low risk
Bias due to deviation from intended intervention (assignment to intervention)	low risk
Bias due to deviation from intended intervention (adhering to intervention)	NA
Bias due to missing outcome data	low risk
Bias in measurement of the outcome	low risk
Bias in selection of the reported result	low risk
Other sources of bias	some concerns
Overall RoB judgment	low risk

Toxicity

Outcome type As specificed by the authors	Primary
Outcome specification	NA
Type of measurement	Observation
Results during intervention - Results during intervention means that the time of data collection is during or shortly after the period of the intervention (e.g. on the last day or a few days after). The results therefore still relate to the direct effects of the intervention.</br>- Results after intervention means there is a longer break between the time of data collection and the end of the intervention, e.g. more than a week. The results relate more to long-term effects.</br>- If a categorization in Results during vs. after intervention is not possible (e.g. survival data), the results are summarized under Results after intervention under the headline "Overall".	NA
Results after intervention - Results during intervention means that the time of data collection is during or shortly after the period of the intervention (e.g. on the last day or a few days after). The results therefore still relate to the direct effects of the intervention.</br>- Results after intervention means there is a longer break between the time of data collection and the end of the intervention, e.g. more than a week. The results relate more to long-term effects.</br>- If a categorization in Results during vs. after intervention is not possible (e.g. survival data), the results are summarized under Results after intervention under the headline "Overall".	Collected in 865 patients in intervention arm and 477 in placebo arm: Grade 1-2 toxicity in intervention arm: 31% and placebo arm: 26% of subjects Grade ≥ 3 toxicity occurred in less than 2% of subjects in intervention arm and 3% in placebo arm n=1 patient in placebo arm had constitutional lethal toxicity no arm comparison performed

Risk of Bias Assessment: Cochrane RoB tool 2.0
Bias arising from the randomization process	low risk
Bias due to deviation from intended intervention (assignment to intervention)	low risk
Bias due to deviation from intended intervention (adhering to intervention)	NA
Bias due to missing outcome data	low risk
Bias in measurement of the outcome	low risk
Bias in selection of the reported result	low risk
Other sources of bias	some concerns
Overall RoB judgment	low risk

Selenium level

Outcome type As specificed by the authors	Others
Outcome specification	NA
Type of measurement	Blood Test
Results during intervention - Results during intervention means that the time of data collection is during or shortly after the period of the intervention (e.g. on the last day or a few days after). The results therefore still relate to the direct effects of the intervention.</br>- Results after intervention means there is a longer break between the time of data collection and the end of the intervention, e.g. more than a week. The results relate more to long-term effects.</br>- If a categorization in Results during vs. after intervention is not possible (e.g. survival data), the results are summarized under Results after intervention under the headline "Overall".	NA
Results after intervention - Results during intervention means that the time of data collection is during or shortly after the period of the intervention (e.g. on the last day or a few days after). The results therefore still relate to the direct effects of the intervention.</br>- Results after intervention means there is a longer break between the time of data collection and the end of the intervention, e.g. more than a week. The results relate more to long-term effects.</br>- If a categorization in Results during vs. after intervention is not possible (e.g. survival data), the results are summarized under Results after intervention under the headline "Overall".	Selenium level at baseline (n=1,022), year 2 (n=375), and year 4 (n=194): at baseline selenium level mostly in normal range, after 2 and 4 years significantly increased selenium concentration in the intervention arm

Risk of Bias Assessment: Cochrane RoB tool 2.0
Bias arising from the randomization process	low risk
Bias due to deviation from intended intervention (assignment to intervention)	low risk
Bias due to deviation from intended intervention (adhering to intervention)	NA
Bias due to missing outcome data	low risk
Bias in measurement of the outcome	low risk
Bias in selection of the reported result	low risk
Other sources of bias	some concerns
Overall RoB judgment	low risk

Funding and Conflicts of Interest

Funding	According to authors none.
Conflicts of Interest	“Employment or Leadership Position: None; Consultant or Advisory Role: David H. Johnson, Peloton Therapeutics (C), Mirna Therapeutics (C) Stock Ownership: None; Honoraria: None; research; Expert Testimony: None; Patents: None Other Remuneration: None“

Further points for assessing the study

Sample

Power analysis performed	Yes
- Sample size corresponds to power analysis	No
- Reasons for insufficient sample size based on power analysis	No
If no power analysis performed: at least moderate sample size (n >= 30 per arm)	NA
Ethnicity mentioned	No

Alternative Explanation

Other explanations for an effect besides the investigated intervention	Yes
- Possibility of attention effects	NA
- Possibility of placebo effects	NA
- Other reasons	Subjects could continue to take supplements that contained ≤70 µg selenium and also others that did not contain selenium

Statistics

Correct use of parametric and non-parametric tests Testing for normal distribution only necessary if parametric tests are used, NI: use of parametric tests without report of normal distribution testing	Yes
Correction for multiple testing	NA
Measurement of compliance	Yes
Consistent reporting in numbers (figures, flowchart, abstract, results)	Yes
Comprehensive and coherent reporting	No
Cross-over	No
- Sufficient washout period	NA
- Tested for carry-over effects	NA
- Tested for sequence effects	NA

Interpretation of results

Effect sizes reported (clinical vs. statistical significance)	NA
Side effects systematically recorded	Yes
Side effects considered in result interpretation	Yes
Ethics votum	NI

Additional Notes

Note: Interim analysis in October 2009 (46% of endpoints reached with 1561 patients randomized), study was stopped by DMC on November 5, 2009 as a trend for a placebo arm benefit was found; June 2011 update with 54% of the endpoints

PRO:

• Prior testing of compliance in a 4-week phase and then testing every 3 months.
• Stratification by demographic variables.
• Power analysis conducted.
• Intent-to-treat analysis performed.
• Measurement of selenium concentration.
• Very detailed presentation of data.

CONTRA:

• Participants could continue taking supplements containing ≤70 µg of selenium and others containing no selenium.
• Subgroup analysis based on selenium concentration level but without significance values.
• No mention of ethics approval.
• Confusing description of DFS (disease-free survival).
• Figure 2b describes OS (overall survival) in days (typographical error).
• OS curves diverge significantly after approximately 7.5 years, but no explanation of these data is provided.