Nej Mo A 1811850

The n e w e ng l a n d j o u r na l of m e dic i n e
Original Article
Randomized Trial of Four Treatment

Approaches for Actinic Keratosis
Maud H.E. Jansen, M.D., Janneke P.H.M. Kessels, M.D., Ph.D.,
Patty J. Nelemans, M.D., Ph.D., Nina Kouloubis, M.D.,
Aimee H.M.M. Arits, M.D., Ph.D., Han P.A. van Pelt, M.D., Ph.D.,
Patricia J.F. Quaedvlieg, M.D., Ph.D., Brigitte A.B. Essers, Ph.D.,
Peter M. Steijlen, M.D., Ph.D., Nicole W.J. Kelleners‑Smeets, M.D., Ph.D.,
and Klara Mosterd, M.D., Ph.D.
A BS T R AC T
BACKGROUND
Actinic keratosis is the most frequent premalignant skin disease in the white From the Departments of Dermatology
population. In current guidelines, no clear recommendations are made about (M.H.E.J., J.P.H.M.K., N.K., A.H.M.M.A.,
P.M.S., N.W.J.K.-S., K.M.) and Clinical
which treatment is preferred. Epidemiology and Medical Technology
Assessment (B.A.B.E.), Maastricht Uni-
METHODS versity Medical Center, and the GROW
We investigated the effectiveness of four frequently used field-directed treatments School for Oncology and Developmental
Biology (M.H.E.J., J.P.H.M.K., A.H.M.M.A.,
(for multiple lesions in a continuous area). Patients with a clinical diagnosis of five P.M.S., N.W.J.K.-S., K.M.) and the De-
or more actinic keratosis lesions on the head, involving one continuous area of partment of Epidemiology (P.J.N.), Maas-
25 to 100 cm2, were enrolled at four Dutch hospitals. Patients were randomly as- tricht University, Maastricht, the Depart-
ment of Dermatology, Zuyderland Medical
signed to treatment with 5% fluorouracil cream, 5% imiquimod cream, methyl Center, Heerlen (J.P.H.M.K., P.J.F.Q.), the
aminolevulinate photodynamic therapy (MAL-PDT), or 0.015% ingenol mebutate Department of Dermatology, Catharina
gel. The primary outcome was the proportion of patients with a reduction of 75% Hospital, Eindhoven (A.H.M.M.A., P.M.S.),
and the Department of Dermatology,
or more in the number of actinic keratosis lesions from baseline to 12 months VieCuri Medical Center, Venlo (H.P.A.P.)
after the end of treatment. Both a modified intention-to-treat analysis and a per- — all in the Netherlands. Address reprint
protocol analysis were performed. requests to Dr. Jansen at the Department
of Dermatology, Maastricht University
RESULTS Medical Center, P. Debyelaan 25, P.O. Box
5800, 6202 AZ Maastricht, the Nether-
A total of 624 patients were included from November 2014 through March 2017. lands, or at maud.jansen@mumc.nl.
At 12 months after the end of treatment, the cumulative probability of remaining
Drs. Jansen and Kessels contributed
free from treatment failure was significantly higher among patients who received equally to this article.
fluorouracil (74.7%; 95% confidence interval [CI], 66.8 to 81.0) than among those
N Engl J Med 2019;380:935-46.
who received imiquimod (53.9%; 95% CI, 45.4 to 61.6), MAL-PDT (37.7%; 95% CI, DOI: 10.1056/NEJMoa1811850
30.0 to 45.3), or ingenol mebutate (28.9%; 95% CI, 21.8 to 36.3). As compared with Copyright © 2019 Massachusetts Medical Society.
fluorouracil, the hazard ratio for treatment failure was 2.03 (95% CI, 1.36 to 3.04)
with imiquimod, 2.73 (95% CI, 1.87 to 3.99) with MAL-PDT, and 3.33 (95% CI,
2.29 to 4.85) with ingenol mebutate (P≤0.001 for all comparisons). No unexpected
toxic effects were documented.
CONCLUSIONS
At 12 months after the end of treatment in patients with multiple actinic keratosis
lesions on the head, 5% fluorouracil cream was the most effective of four field-
directed treatments. (Funded by the Netherlands Organization for Health Research
and Development; ClinicalTrials.gov number, NCT02281682.)
n engl j med 380;10 nejm.org March 7, 2019 935

The New England Journal of Medicine
Downloaded from nejm.org on October 20, 2022. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
A
ctinic keratosis is the most fre- of four hospitals in the Netherlands. The trial
quent premalignant skin disease in the was performed according to the principles of the
white population and is caused by expo- Declaration of Helsinki, and the protocol (avail-
sure to ultraviolet radiation. With a prevalence of able with the full text of this article at NEJM.org)
37.5% among whites 50 years of age or older, was approved by the local medical ethics com-
actinic keratosis is one of the most frequent mittee. No commercial support was provided for
reasons for patients to visit a dermatologist.1-3 If the trial. The trial drugs were purchased as a
left untreated, actinic keratosis may develop into part of routine care of the patients. No company
squamous-cell carcinoma.4,5 However, no defin- had any role in the design of the trial, the col-
able clinical characteristics distinguish which lection or analysis of the data, or the writing of
actinic keratosis lesions pose a risk and what the manuscript. The authors vouch for the com-
proportion of actinic keratosis lesions will prog- pleteness and accuracy of the data and analyses
ress into a carcinoma. Percentages that have and for the fidelity of the trial to the protocol.
been reported in studies range from 0.025 to Patients 18 years of age or older with a clini-
16% per actinic keratosis lesion per year.6-9 cal diagnosis of five or more actinic keratosis
The recurrence rate after treatment is high, lesions in one continuous area of skin measur-
often leading to repetitive treatments, although ing 25 to 100 cm2 in the head and neck area
research suggests that the effect of fluorouracil were eligible for participation. Clinical diagnosis
(also called 5-fluorouracil) on actinic keratosis of actinic keratosis was made by the patient’s
reduction can last for years.10,11 Solitary lesions own physician and verified by one of two inves-
can be treated with cryotherapy. However, pa- tigators who assessed the end points without
tients with actinic keratosis often present with knowing the treatment assignment. In the case
multiple lesions in one continuous area (so-called of a larger affected area, the investigator select-
field change). Generally, field-directed therapies ed the area with the most pronounced lesions.
are preferred, because they not only are thera- All grades of actinic keratosis lesions (Olsen
peutically effective for present actinic keratosis grades I to III, a three-point grading system
but also may have a prophylactic effect on the based on thickness of hyperkeratosis, with higher
development of new lesions; in addition, they grades indicating more severe lesions) were in-
may prevent the development of squamous-cell cluded. (For details on the Olsen scale, see the
carcinoma.11-13 Supplementary Appendix, available at NEJM.org.)
Current guidelines provide no clear recom- Patients were not eligible to participate if they
mendations about which treatment approach is had received any treatment for actinic keratosis
preferred.14-17 Frequently prescribed and studied (including cryotherapy) in the target area or had
field-directed treatment approaches are fluoro- used systemic retinoids or systemic immunosup-
uracil cream, imiquimod cream, photodynamic pressant drugs within 3 months before inclusion.
therapy (PDT), and ingenol mebutate gel. Other reasons for exclusion were suspicion of
Currently, the choice of treatment often de- cancer in the target area, porphyria, allergy to
pends on the preferences of patients and their trial drugs, pregnancy or breast-feeding, or a per-
treating physicians. Evidence from randomized sonal history of a genetic skin-cancer disorder.
trials with direct comparison between treatments All patients provided written informed consent
and with long-term follow-up is scarce.11 The before randomization.
aim of this randomized, controlled trial was to
compare treatment success at 12 months of 5% Randomization, Treatment, and Assessments
fluorouracil cream, 5% imiquimod cream, methyl Patients were randomly assigned to one of four
aminolevulinate PDT (MAL-PDT), and 0.015% investigated treatments in a 1:1:1:1 ratio. Random-
ingenol mebutate gel in patients with actinic ization lists that were based on minimization
keratosis lesions of any grade. were computer generated with Alea software.18
Stratifying factors were treatment center and
severity of actinic keratosis. An investigator who
Me thods
was aware of the treatment assignments (the
Trial Design and Population second author) performed the randomization
This multicenter, single-blind, randomized trial procedure. The second author was also respon-
was conducted at the dermatology departments sible for the distribution of trial information and
936 n engl j med 380;10 nejm.org March 7, 2019

Four Treatment Approaches for Actinic Ker atosis
medication and managed and recorded the ad- tion from baseline in the number of actinic kera-
verse events as well as adherence. An investigator tosis lesions), adverse events, adherence, patient
who was unaware of the treatment assignments satisfaction with treatment, health-related qual-
(the first author) evaluated all trial end points, ity of life, and cosmetic results. Details are pro-
with the exception of adverse events and adher- vided in the Supplementary Appendix.
ence during the baseline and follow-up visits.
The first author determined the number and Statistical Analysis
extent of actinic keratosis lesions at the baseline The sample size of this trial was based on the
visit and at 3 and 12 months after the end of primary end point, lesion reduction of at least
treatment. All lesions were drawn with their 75% at 12 months after the end of treatment. On
exact location on a transparent sheet with the the basis of previous studies, we estimated that
use of physical reference points as landmarks. 65% of patients would have lesion reduction of
The severity of each lesion was graded with the at least 75% at 12 months after the end of treat-
Olsen scale.19 Owing to the nature of the trial ment.20 To enable detection of an absolute differ-
medication, patients were aware of the treatment ence of 15 percentage points between treatment
assignments. groups with a power of 80% and an alpha level
Details about the treatment protocols of of 5%, a total of 140 patients were required per
fluorouracil, imiquimod, ingenol mebutate, and treatment group. To account for a potential loss
MAL-PDT are provided in the Supplementary to follow-up of 10%, a total of 624 (4 × 156) pa-
Appendix. In all patients, superficial curettage tients needed to be included.
of all actinic keratosis lesions was performed The cumulative probability of remaining free
manually before every treatment or retreatment, from treatment failure at 12 months after the
and patients did not receive anesthesia. end of treatment was calculated with the use of
The treatment strategy entailed a first treat- Kaplan–Meier survival analysis of the observed
ment and allowed for a retreatment in case of in- outcome data at 3 and 12 months. Survival
sufficient treatment response, defined as a lesion analysis was used to account for patients who
response of less than 75% at the first follow-up were lost to follow-up and whose data were cen-
visit (Fig. 1). For patients assigned to receive sored at the date of the last follow-up visit.
fluorouracil, ingenol mebutate, or MAL-PDT, ini- Cox regression analysis was used to calculate
tial treatment response was evaluated 3 months hazard ratios for treatment failure with 95% con-
after the first treatment. For patients assigned to fidence intervals and P values; the treatment
receive imiquimod, initial response was evalu- group with the highest rate of success was used
ated 1 month after the last treatment day, in as the reference group. Variables with dichoto-
accordance with the summary-of-product-char- mous outcomes were compared between the treat-
acteristics guideline for imiquimod. All patients ment groups with the use of the chi-square test
could receive a maximum of two courses of the or Fisher’s exact test for proportions. For continu-
assigned treatment. In case of less than 75% ous variables, between-group differences were
clearance of actinic keratosis 3 months after the compared with the use of analysis of variance (if
final treatment, those patients were assessed as normally distributed) or a nonparametric test for
having treatment failure for the final analysis. independent samples (if not normally distributed).
A Bonferroni adjustment was made for multi-
Outcomes ple comparisons with 0.008 (0.05 ÷ 6) as the alpha
The primary outcome of this trial was the pro- value, because there were six possible pairwise
portion of patients who remained free from comparisons between treatment groups. Analyses
treatment failure during 12 months of follow-up were performed with the use of SPSS software,
after the last treatment. Treatment failure was version 23.0 (IBM), and Stata software, version
defined as a reduction of less than 75% in the 14.0 (StataCorp).
number of actinic keratosis lesions counted at
baseline and could occur at 3 months after the R e sult s
last treatment (initial failure) or at 12 months
after initial successful treatment. Secondary out- Trial Population
comes were initial treatment success at 3 months From November 2014 through March 2017, a total
after the last treatment (defined as ≥75% reduc- of 1174 patients were assessed for eligibility

938
Evaluation: Evaluation: Evaluation: Evaluation:
1 Mo after End 3 Mo after End 3 Mo after End 12 Mo after End
Diagnosis Treatment of Treatment of Treatment of Retreatment of Treatment
Success
Success
Failure
Fluorouracil
Success
Retreatment
Success
Failure Failure
Failure
Retreatment
Success
Success
Failure
Success
Success
Retreatment
The
Success
Failure Failure
Imiquimod Failure
Retreatment
Success
Consultation Retreatment Success
Failure Failure
Retreatment Failure
Success
Success
Failure
MAL-PDT
Success
Retreatment
n e w e ng l a n d j o u r na l
Success

of
Failure Failure
Failure
Retreatment
Success
n engl j med 380;10 nejm.org March 7, 2019

Success
Failure

m e dic i n e
Ingenol mebutate
Success
Retreatment
Success
Failure Failure
Failure
Retreatment
Secondary Primary
Key: Success Success
end point end point
Insufficient response, Primary
Failure Failure
retreatment necessary end point
1174 Patients were assessed for eligibility
550 Declined to participate
624 Underwent randomization
155 Were assigned to 156 Were assigned to 156 Were assigned to 157 Were assigned to
receive fluorouracil receive imiquimod receive MAL-PDT receive ingenol mebutate
4 Did not begin treat- 3 Did not begin treat-

ment ment 1 Did not begin treat- 6 Did not begin treat-
2 Were lost to follow-up 4 Were lost to follow-up ment ment
1 Died 1 Died 1 Withdrew 1 Withdrew
1 Withdrew 3 Withdrew
149 Were included in the 149 Were included in the 154 Were included in the 150 Were included in the
3-mo follow-up 3-mo follow-up 3-mo follow-up 3-mo follow-up
14 Had treatment failure 37 Had treatment failure 49 Had treatment failure

36 Had treatment failure
4 Were lost to follow-up 2 Were lost to follow-up 3 Were lost to follow-up
5 Withdrew
1 Died 1 Died 1 Died
3 Withdrew 1 Withdrew 2 Withdrew
131 Were included in the 108 Were included in the 115 Were included in the 98 Were included in the
12-mo follow-up 12-mo follow-up 12-mo follow-up 12-mo follow-up
23 Had treatment failure 31 Had treatment failure 58 Had treatment failure 56 Had treatment failure
Figure 2. Screening, Randomization, and Follow-up.
(Fig. 2). Of those, 550 patients declined to par- tained, and 38 patients did not give a reason for
ticipate for the following reasons: preference or declining to participate.
disfavor regarding one or more of the studied A total of 624 patients underwent randomiza-
treatments (197 patients), old age or coexisting tion in four hospitals: Maastricht University Med-
conditions (113), disapproval to receive a treat- ical Center (247 patients), Catharina Hospital
ment by randomization (88), decision of the pa- (176), VieCuri Medical Center (108), and Zuyder-
tient not to have actinic keratosis treated (53), land Medical Center (93). A total of 155 patients
logistic reasons (24), concern about possible side were randomly assigned to fluorouracil, 156 to
effects (24), treatment costs (9), and preference imiquimod, 156 to MAL-PDT, and 157 to ingenol
for treatment in a different hospital (3). One pa- mebutate. A total of 14 patients did not start
tient died before informed consent could be ob- treatment, and 8 patients were treated but did
not attend the 3-month follow-up visit. Between
3 and 12 months, 14 patients were lost to follow-
Figure 1 (facing page). Schematic Presentation
of the Treatment Strategies.
up (Fig. 2). Eight crossovers occurred before the
MAL-PDT denotes methyl aminolevulinate photo
assigned treatment was started, all because pa-
dynamic therapy. tients preferred a different therapy: 1 patient as-
signed to fluorouracil received MAL-PDT; 1 patient

assigned to imiquimod received fluorouracil; 5 A treatment failure after one treatment cycle
patients assigned to MAL-PDT received fluoro- was observed in 14.8% of the patients (23 of
uracil (3) or ingenol mebutate (2); and 1 patient 155) after fluorouracil therapy, 37.2% (58 of 156)
assigned to ingenol mebutate received fluoroura- after imiquimod therapy, 34.6% (54 of 156) after
cil. No substantial imbalances in baseline char- MAL-PDT, and 47.8% (75 of 157) after ingenol
acteristics were observed between treatment mebutate therapy. According to the trial proto-
groups (Table 1). Although patients with actinic col, a second treatment cycle was offered to all
keratosis lesions on the neck were eligible for patients with treatment failure after one cycle.
the trial, all the patients had lesions on the head However, some patients declined a second treat-
only (vertex or face). ment. This occurred more frequently in the imi
quimod, MAL-PDT, and ingenol mebutate groups
Effectiveness than in the fluorouracil group. Retreatment oc-
A modified intention-to-treat analysis was based curred in 19 of 23 patients (83%) with treatment
on 602 randomly assigned patients who started failure after fluorouracil therapy. These percent-
treatment and for whom data regarding the pri- ages were lower in the other groups: 44 of 58
mary outcome were available. A total of 14 pa- patients (76%) in the imiquimod group (P = 0.57),
tients who declined treatment after randomiza- 41 of 54 patients (76%) in the MAL-PDT group
tion and 8 patients with early loss to follow-up (P = 0.77), and 60 of 75 patients (80%) in the
were excluded (Table S1 in the Supplementary ingenol mebutate group (P = 1.00).
Appendix). In the modified intention-to-treat When we restricted the modified intention-
analysis, data were analyzed according to the to-treat analysis to patients with grade I or II
treatment to which the patient was assigned by actinic keratosis lesions, the percentages with
randomization. treatment success were similar to those in the
The cumulative probability of treatment suc- unrestricted analysis, and fluorouracil remained
cess for fluorouracil was 74.7% (95% confidence superior to the other treatments. For fluoroura-
interval [CI], 66.8 to 81.0). For imiquimod, MAL- cil, the percentage was 75.3% (95% CI, 67.2 to
PDT, and ingenol mebutate, these percentages 81.7). For imiquimod, PDT, and ingenol mebu-
were 53.9% (95% CI, 45.4 to 61.6), 37.7% (95% tate, the percentages were 52.6% (95% CI, 43.7
CI, 30.0 to 45.3), and 28.9% (95% CI, 21.8 to to 60.7), 38.7% (95% CI, 30.7 to 46.7), and
36.3), respectively, according to the modified 30.2% (95% CI, 22.6 to 38.1), respectively. The
intention-to-treat analysis (Table 2). A Bonfer- group with grade III actinic keratosis lesions
roni adjustment was made for multiple compari- was too small for separate analysis.
sons with the use of an alpha of 0.008 (0.05 ÷ 6),
and the differences between fluorouracil cream Adverse Events
and imiquimod, PDT, and ingenol mebutate Data on adverse events (from completed patient
were significant. diaries) were available for 135 patients who re-
A per-protocol analysis was based on 555 ceived fluorouracil, 121 who received imiquimod,
patients who were treated and had full adher- 117 who received MAL-PDT, and 140 who re-
ence to the treatment protocol. A total of 46 ceived ingenol mebutate. There were no serious
patients with initial treatment failure who de- adverse events that were considered by the in-
clined retreatment were excluded. One other vestigators and the medical ethics committee
patient who requested retreatment with a differ- to be related to the trial treatment. Table 3
ent therapy was also excluded (Table S1 in the shows the percentages of patients who reported
Supplementary Appendix). In the per-protocol adverse events during treatment or the 2 weeks
analysis, data on the 8 patients who crossed over after the end of treatment. No patients discon-
were analyzed according to the treatment they tinued the trial because of adverse events.
actually received. For all treatments, the proba-
bility of remaining free from treatment failure Other Secondary Outcomes
was slightly higher in the per-protocol popula- The percentage of patients with 100% adherence
tion than in the modified intention-to-treat was higher in the ingenol mebutate group
population, but the differences between fluoro- (98.7%) and the MAL-PDT group (96.8%) than in
uracil cream and the other treatments remained the fluorouracil group (88.7%) and the imiquimod
significant (Table 2). group (88.2%). Patient satisfaction with treatment

Table 1. Baseline Characteristics of the Patients in the Modified Intention-to-Treat Population.*
Ingenol
Total Fluorouracil Imiquimod MAL-PDT Mebutate
Characteristic (N = 624) (N = 155) (N = 156) (N = 156) (N = 157)
Sex — no. (%)
Male 558 (89.4) 136 (87.7) 143 (91.7) 140 (89.7) 139 (88.5)
Female 66 (10.6) 19 (12.3) 13 (8.3) 16 (10.3) 18 (11.5)
Median age (range) — yr 73 (48–94) 74 (48–90) 73 (59–89) 73 (55–90) 72 (51–94)
Skin type — no. (%)†
I 245 (39.3) 63 (40.6) 67 (42.9) 54 (34.6) 61 (38.9)
II 333 (53.4) 81 (52.3) 79 (50.6) 92 (59.0) 81 (51.6)
III 46 (7.4) 11 (7.1) 10 (6.4) 10 (6.4) 15 (9.6)
History of actinic keratosis — no. (%)
Yes 487 (78.0) 121 (78.1) 129 (82.7) 115 (73.7) 122 (77.7)
No 137 (22.0) 34 (21.9) 27 (17.3) 41 (26.3) 35 (22.3)
History of nonmelanoma skin cancer — no. (%)‡
Yes 353 (56.6) 90 (58.1) 82 (52.6) 86 (55.1) 95 (60.5)
No 271 (43.4) 65 (41.9) 74 (47.4) 70 (44.9) 62 (39.5)
Sun exposure — no. (%)§
Mild 19 (3.0) 6 (3.9) 5 (3.2) 5 (3.2) 3 (1.9)
Moderate 283 (45.4) 69 (44.5) 73 (46.8) 72 (46.2) 69 (43.9)
Severe 322 (51.6) 80 (51.6) 78 (50.0) 79 (50.6) 85 (54.1)
History of immunosuppressive therapy >3 mo
before inclusion — no. (%)¶
Yes 84 (13.5) 18 (11.6) 25 (16.0) 19 (12.2) 22 (14.0)
No 540 (86.5) 137 (88.4) 131 (84.0) 137 (87.8) 135 (86.0)
Median treated area (range) — cm2 81 (25–100) 80 (27–100) 86.5 (25–100) 81 (25–100) 78 (25–100)
Median no. of actinic keratosis lesions (range) 16 (5–48) 16 (5–48) 16.5 (5–37) 16 (5–38) 15 (5–40)
Severity of actinic keratosis — no. (%)‖
Olsen grade I or II lesions 575 (92.1) 144 (92.9) 143 (91.7) 144 (92.3) 144 (91.7)
≥1 Lesion of Olsen grade III 49 (7.9) 11 (7.1) 13 (8.3) 12 (7.7) 13 (8.3)
Location of lesions — no. (%)
Vertex 321 (51.4) 78 (50.3) 78 (50.0) 80 (51.3) 85 (54.1)
Face 303 (48.6) 77 (49.7) 78 (50.0) 76 (48.7) 72 (45.9)
Trial site — no. (%)**
Maastricht 247 (39.6) 61 (39.4) 62 (39.7) 62 (39.7) 62 (39.5)
Eindhoven 176 (28.2) 43 (27.7) 44 (28.2) 44 (28.2) 45 (28.7)
Venlo 108 (17.3) 27 (17.4) 27 (17.3) 27 (17.3) 27 (17.2)
Heerlen 93 (14.9) 24 (15.5) 23 (14.7) 23 (14.7) 23 (14.6)
* There were no significant between-group differences in the characteristics listed here. Percentages may not total 100 because of rounding.
MAL-PDT denotes methyl aminolevulinate photodynamic therapy.
† Skin type was graded according to Fitzpatrick’s classification. Type I indicates always burns, never tans. Type II indicates burns easily, tans
minimally. Type III indicates sometimes burns, slowly tans to light brown.
‡ Nonmelanoma skin cancer was defined as cutaneous melanoma or keratinocyte cancer.
§ Mild exposure was defined as no history of frequent sun exposure due to profession, tanning, or hobbies. Moderate exposure was defined
as sun exposure only during leisure time or holidays. Severe exposure was defined as outdoor profession, frequent use of solar beds, a
history of living in tropical areas, or participation in water sports.
¶ Immunosuppressive drugs included prednisolone, methotrexate, azathioprine, tacrolimus, or biologic agents.
‖ The severity of each lesion was graded with the Olsen scale.19 This three-point grading system is based on thickness of hyperkeratosis,
with higher grades indicating more severe lesions.
** The trial was conducted at four hospitals in the Netherlands: Maastricht University Medical Center (Maastricht), Catharina Hospital
(Eindhoven), VieCuri Medical Center (Venlo), and Zuyderland Medical Center (Heerlen).

and increase in health-related quality of life were
* Because the fluorouracil group had the highest rate of treatment success, it was used as the reference group, according to the statistical analysis plan. CI denotes confidence interval.
P Value‡
highest in the fluorouracil group. Good-to-excel-
0.001
<0.001
<0.001
0.001
<0.001
<0.001
lent cosmetic outcome was more often observed
in the MAL-PDT group (96.6%) and the ingenol
mebutate group (95.1%) than in the fluorouracil
group (90.3%) and the imiquimod group (89.7%).
Hazard Ratio (95% CI)
Detailed results are provided in the Supplemen-

2.03 (1.36–3.04)
2.73 (1.87–3.99)
3.33 (2.29–4.85)
3.33 (2.25–4.94)
2.03 (1.33–3.10)
2.63 (1.76–3.94)
tary Appendix.
1.00
1.00
Discussion
This trial showed that 5% fluorouracil was signifi-
cantly more effective than imiquimod, MAL-PDT,
Cumulative Probability of Remaining Free from
Table 2. Cumulative Probability of Treatment Success at 3 and 12 Months after the End of Treatment and Hazard Ratios for Treatment Failure.*
During 12 Mo after
End of Treatment
74.7 (66.8–81.0)
53.9 (45.4–61.6)
37.7 (30.0–45.3)
28.9 (21.8–36.3)
76.4 (68.6–82.5)
56.7 (47.7–64.7)
42.4 (33.9–50.5)
31.8 (24.1–39.8)
or ingenol mebutate at 12 months after the end
of treatment for multiple actinic keratosis lesions
Treatment Failure (95% CI)†
in a continuous area. Findings from the modi-

fied intention-to-treat analysis and the per-pro-
tocol analysis were similar, which indicates the
percent
robustness of the results.

Although there is substantial literature about
During 3 Mo after
End of Treatment
90.6 (84.7–94.3)
76.0 (68.4–82.0)
67.3 (59.2–74.2)
93.2 (87.7–96.3)
80.7 (73.0–86.5)
83.2 (75.8–88.5)
75.8 (68.1–81.9)
75.0 (66.8–81.4)
different field-directed and lesion-directed treat-

ments, studies often lack head-to-head compari-
sons, differ substantially in choice of outcome
measures, and are underpowered.14,21-23 Our
randomized clinical trial compared four field-
directed treatments with 12 months of follow-up.
Previously, two network meta-analyses have
End of Treatment
108/131 (82.4)
76/107 (71.0)
57/115 (49.6)
109/133 (82.0)
73/104 (70.2)
57/112 (50.9)
been published. One meta-analyis, by Vegter and

42/98 (42.9)
42/99 (42.4)
12 Mo after
Tolley,24 indicated that aminolevulinic acid PDT

number/total number (percent)
(ALA-PDT) with the use of BF-200 ALA gel re-

Treatment Success
sulted in the highest probability (75.8%) of total

clearance of actinic keratosis lesions, as com-
pared with 0.5% fluorouracil (59.9%), 5% imi
quimod (56.3%), and MAL-PDT (54.8%), with a
End of Treatment
135/149 (90.6)
113/149 (75.8)
117/154 (76.0)
101/150 (67.3)
137/147 (93.2)
109/135 (80.7)
114/137 (83.2)
102/136 (75.0)
† Cumulative probabilities were based on Kaplan–Meier analysis.
follow-up of 3 to 12 months. But this meta-

3 Mo after
analysis did not include 5% fluorouracil, as was

used in our trial. The other meta-analysis, by
Gupta and Paquet,25 suggested that 5% fluoro-
uracil was the most effective treatment when
‡ P values were based on Cox regression analysis.
complete clearance of all lesions was assessed.

However, in the 2015 European Dermatology
Modified intention-to-treat population
Forum guidelines, the majority of experts did not

express a preference for any of the most com-
monly prescribed treatments.16 They agreed that
3.75% imiquimod, ALA-PDT, MAL-PDT, ingenol
Per-protocol population
mebutate (0.015% or 0.050%), and 0.5% fluoro-

uracil were equally effective in patients with
Ingenol mebutate
Ingenol mebutate
multiple actinic keratosis lesions.16 However,

there was less agreement on the effectiveness of
Fluorouracil
Fluorouracil
Imiquimod
Imiquimod
MAL-PDT
MAL-PDT
5% fluorouracil. In our trial, we used the most

Variable
commonly prescribed dosing regimens of the

therapies studied. Alternative regimens (e.g., dif-

ferent concentrations or duration of therapy) puts a high burden on the health care system,
might result in differences in effectiveness be- with 5 million dermatology visits per year in the
tween treatments. United States alone.26,27 Our results could affect
An important gap in the current literature is treatment choices in both dermatology and pri-
that most studies assessing the effectiveness of mary care. From a cost perspective, fluorouracil
field-directed treatments exclude grade III ac- is also the most attractive option.28 It is expected
tinic keratosis lesions. In this trial, the popula- that a substantial cost reduction could be
tion included patients with grade III actinic achieved with more uniformity in care and the
keratosis lesions; in this way, it is more repre- choice for effective therapy.
sentative of patients seen in daily practice. Ex- This randomized trial has some limitations.
clusion of patients with grade III lesions was Approximately half the patients who were as-
associated with slightly higher rates of success sessed for eligibility declined to participate in
in the fluorouracil, MAL-PDT, and ingenol meb- this trial, usually because of personal preference
utate groups than the rates in the unrestricted or disfavor regarding a specific therapy. Patients’
analysis. declining to participate is a common problem in
The reported adverse events in this trial are randomized trials and may threaten external valid-
well-known treatment-related side effects that ity. The median age of the eligible trial popula-
have been described in the corresponding sum- tion, 75 years, was similar to that of patients
mary of product characteristics. No treatment- who participated, but the ratio of men to women
related serious adverse events occurred. Overall, was 4:1 in the eligible population and 9:1 in the
treatment with fluorouracil was not associated trial population, which suggests that men were
with a higher frequency of adverse events during more willing to participate. Generalizability of
and after treatment than the other treatments. the findings would be affected if the effective-
High scores for pain and burning sensation were ness of the evaluated treatments depends on sex,
reported most often during MAL-PDT treatment. but sex-specific treatment response seems unlike
Pain can be an important reason for a patient to ly. To avoid substantial interobserver variation,
decline further treatment. In our trial, only 3.2% all counts were performed by a single observer
of the patients (5 of 155) in the MAL-PDT group who was unaware of the treatment assignments.
did not complete the entire treatment owing to There may still be intraobserver variation, but ran-
pain, but the proportion of patients who would dom measurement errors result in nondifferen-
undergo this treatment again and would recom- tial misclassification, which tends to dilute dif-
mend it to others was lower than for the other ferences between baseline and follow-up counts
treatments, which indicates that pain may have of actinic keratosis lesions. However, potential
influenced patient satisfaction with PDT. Satis- underestimation of the reduction in actinic kera-
faction with treatment and improvement in tosis lesions will occur in all treatment groups
health-related quality of life at 12 months after and is unlikely to affect the comparison among
the end of treatment were highest in the fluoro- groups.29 Adherence to therapy with fluorouracil
uracil group. This may be explained in part by and imiquimod, which are applied for 4 con-
the fact that fluorouracil was the most effective secutive weeks, was high in this trial (88.7% and
treatment. However, the high proportion of pa- 88.2%, respectively), but in daily practice, adher-
tients willing to undergo retreatment after ini- ence may be lower. In this respect, ingenol
tial treatment failure also suggests that patients mebutate, which is applied for only 3 consecu-
treated with fluorouracil may have had less in- tive days, had the advantage of better adherence
convenience and discomfort than those treated (98.7%), but owing to the observed low probabil-
with imiquimod, MAL-PDT, or ingenol mebu- ity of remaining free from treatment failure of
tate, for which the proportion of patients who only 28.9% in the long term, ingenol mebutate
declined retreatment were higher. might be reserved for situations in which alter-
Dermatologists and primary health care pro- native treatments are not feasible.
viders are both confronted with actinic keratosis In our trial, adherence was assessed by ask-
lesions very frequently. Because of the increasing ing patients 2 weeks after the end of treatment
age of the general population and the high re- how often they had used the therapy. This
currence rate of actinic keratosis, this condition method may be subject to error.

Table 3. Adverse Events.*
Ingenol
Fluorouracil Imiquimod MAL-PDT Mebutate
Event (N = 135) (N = 121) (N = 117) (N = 140) P Value†
number (percent)
Any adverse event 125 (92.6) 103 (85.1) 113 (96.6) 134 (95.7) 0.004‡
During treatment
Erythema NA
Moderate or severe 110 (81.5) 88 (72.7) 105 (75.0) 0.22
Absent or mild 25 (18.5) 33 (27.3) 35 (25.0)
Swelling NA
Absent or mild 94 (69.6) 68 (56.2) 81 (57.9)
Erosion NA
Absent or mild 81 (60.0) 63 (52.1) 98 (70.0)
Crusts NA
Moderate or severe 77 (57.0) 83 (68.6) 53 (37.9) <0.001‡
Absent or mild 58 (43.0) 38 (31.4) 87 (62.1)
Vesicles or bullae NA
Moderate or severe 33 (24.4) 38 (31.4) 59 (42.1) 0.007‡
Absent or mild 102 (75.6) 83 (68.6) 81 (57.9)
Scaling NA
Absent or mild 75 (55.6) 70 (57.9) 90 (64.3)
Itching NA
Moderate or severe 84 (62.2) 74 (61.2) 58 (41.4) 0.001‡
Absent or mild 51 (37.8) 47 (38.8) 82 (58.6)
Pain
Severe 22 (16.3) 11 (9.1) 73 (62.4) 17 (12.1) <0.001‡
Moderate 21 (15.6) 21 (17.4) 20 (17.1) 40 (28.6)
Absent or mild 92 (68.1) 89 (73.6) 24 (20.5) 83 (59.3)
Burning sensation
Severe 29 (21.5) 12 (9.9) 78 (66.7) 30 (21.4) <0.001‡
Moderate 34 (25.2) 30 (24.8) 22 (18.8) 42 (30.0)
Absent or mild 72 (53.3) 79 (65.3) 17 (14.5) 68 (48.6)
2 Wk after end of treatment
Erythema
Moderate or severe 79 (58.5) 61 (50.4) 87 (74.4) 65 (46.4) <0.001‡
Absent or mild 56 (41.5) 60 (49.6) 30 (25.6) 75 (53.6)
Swelling
Moderate or severe 31 (23.0) 26 (21.5) 29 (24.8) 41 (29.3) 0.48
Absent or mild 104 (77.0) 95 (78.5) 88 (75.2) 99 (70.7)

Table 3. (Continued.)
Ingenol
Fluorouracil Imiquimod MAL-PDT Mebutate
Event (N = 135) (N = 121) (N = 117) (N = 140) P Value†
number (percent)
Erosion
Absent or mild 86 (63.7) 85 (70.2) 87 (74.4) 110 (78.6)
Crusts
Moderate or severe 66 (48.9) 68 (56.2) 49 (41.9) 87 (62.1) 0.008‡
Absent or mild 69 (51.1) 53 (43.8) 68 (58.1) 53 (37.9)
Vesicles or bullae
Absent or mild 107 (79.3) 104 (86.0) 95 (81.2) 105 (75.0)
Scaling
Moderate or severe 77 (57.0) 46 (38.0) 70 (59.8) 93 (66.4) <0.001‡
Absent or mild 58 (43.0) 75 (62.0) 47 (40.2) 47 (33.6)
Itching
Moderate or severe 75 (55.6) 47 (38.8) 56 (47.9) 85 (60.7) 0.003‡
Absent or mild 60 (44.4) 74 (61.2) 61 (52.1) 55 (39.3)
Pain
Severe 9 (6.7) 7 (5.8) 12 (10.3) 10 (7.1) 0.09
Moderate 15 (11.1) 9 (7.4) 21 (17.9) 24 (17.1)
Absent or mild 111 (82.2) 105 (86.8) 84 (71.8) 106 (75.7)
Burning sensation
Severe 19 (14.1) 5 (4.1) 15 (12.8) 8 (5.7) 0.01
Moderate 18 (13.3) 14 (11.6) 17 (14.5) 32 (22.9)
Absent or mild 98 (72.6) 102 (84.3) 85 (72.6) 100 (71.4)
* Percentages may not total 100 because of rounding. NA denotes not applicable.
† All P values are for comparisons across all four treatment groups.
‡ A P value of 0.008 or less was considered to indicate statistical significance.
In conclusion, we found that after 12 months Dr. Jansen and Dr. Kessels report receiving conference costs
sponsored by Galderma; Dr. Quaedvlieg, receiving advisory
of follow-up, 5% fluorouracil cream was signifi- board fees from LEO Pharma; Dr. Kelleners-Smeets, receiving
cantly more effective than 5% imiquimod cream, advisory board fees from LEO Pharma and conference costs
MAL-PDT, or 0.015% ingenol mebutate gel in sponsored by Galderma; and Dr. Mosterd, receiving trial sup-
plies from Meda. No other potential conflict of interest relevant
the treatment of patients with multiple grade I to this article was reported.
to III actinic keratosis lesions on the head. No Disclosure forms provided by the authors are available with
new toxic effects were identified in this trial. the full text of this article at NEJM.org.
We thank the patients who agreed to participate in this trial;
A data sharing statement provided by the authors is available all nurse practitioners, nursing staff, and employees of the
with the full text of this article at NEJM.org. administrative departments of the participating hospitals; and
Supported by a grant (80-83600-98-3054) from the Nether- the following persons for their efforts on behalf of our trial:
lands Organization for Health Research and Development Drs. S. Dodemont, L. Voeten, M. Hacking, J. Clabbers, N. Ramak-
(ZonMw), a governmental institution that finances research to ers, M. Maris, E. van Loo, J. Havens, and S. Ahmady and Mrs.
improve health care in the Netherlands. A. Ebus.

References
1. Flohil SC, van der Leest RJ, Dowlat a randomized clinical trial. JAMA Derma- 20. Lebwohl M, Dinehart S, Whiting D, et
shahi EA, Hofman A, de Vries E, Nijsten tol 2015;151:952-60. al. Imiquimod 5% cream for the treatment
T. Prevalence of actinic keratosis and its 12. Weinstock MA, Thwin SS, Siegel JA, of actinic keratosis: results from two
risk factors in the general population: the et al. Chemoprevention of basal and squa- phase III, randomized, double-blind, par-
Rotterdam Study. J Invest Dermatol 2013; mous cell carcinoma with a single course allel group, vehicle-controlled trials. J Am
133:1971-8. of fluorouracil, 5%, cream: a randomized Acad Dermatol 2004;50:714-21.
2. Spencer JM, Hazan C, Hsiung SH, clinical trial. JAMA Dermatol 2018;154: 21. Dirschka T, Gupta G, Micali G, et al.
Robins P. Therapeutic decision making 167-74. Real-world approach to actinic keratosis
in the therapy of actinic keratoses. J Drugs 13. Neugebauer R, Levandoski KA, Zhu Z, management: practical treatment algo-
Dermatol 2005;4:296-301. et al. A real-world, community-based co- rithm for office-based dermatology. J Der-
3. Siegel JA, Korgavkar K, Weinstock MA. hort study comparing the effectiveness of matolog Treat 2017;28:431-42.
Current perspective on actinic keratosis: topical fluorouracil versus topical imi 22. Stockfleth E, Ferrandiz C, Grob JJ,
a review. Br J Dermatol 2017;177:350-8. quimod for the treatment of actinic kera- Leigh I, Pehamberger H, Kerl H. Develop-
4. Dinehart SM, Nelson-Adesokan P, tosis. J Am Acad Dermatol 2018;78:710-6. ment of a treatment algorithm for actinic
Cockerell C, Russell S, Brown R. Meta- 14. de Berker D, McGregor JM, Mohd Mu- keratoses: a European Consensus. Eur J
static cutaneous squamous cell carcinoma stapa MF, Exton LS, Hughes BR. British Dermatol 2008;18:651-9.
derived from actinic keratosis. Cancer Association of Dermatologists’ guidelines 23. Stockfleth E, Kerl H. Guidelines for
1997;79:920-3. for the care of patients with actinic kera- the management of actinic keratoses. Eur
5. Ackerman AB, Mones JM. Solar (actin- tosis 2017. Br J Dermatol 2017;176:20-43. J Dermatol 2006;16:599-606.
ic) keratosis is squamous cell carcinoma. 15. Beljaards RC, van der Sande A. Update 24. Vegter S, Tolley K. A network meta-
Br J Dermatol 2006;155:9-22. richtlijn actinische keratosen 2017. Neder- analysis of the relative efficacy of treat-
6. Marks R, Rennie G, Selwood TS. Ma- lands Tijdschrift voor Dermatologie en ments for actinic keratosis of the face or
lignant transformation of solar keratoses Venereologie 2017;27:190-2. scalp in Europe. PLoS One 2014; 9(6):
to squamous cell carcinoma. Lancet 1988; 16. Werner RN, Stockfleth E, Connolly SM, e96829.
1:795-7. et al. Evidence- and consensus-based (S3) 25. Gupta AK, Paquet M. Network meta-
7. Lanoue J, Chen C, Goldenberg G. Ac- guidelines for the treatment of actinic analysis of the outcome ‘participant com-
tinic keratosis as a marker of field cancer- keratosis — International League of Der- plete clearance’ in nonimmunosuppressed
ization in excision specimens of cutane- matological Societies in cooperation with participants of eight interventions for ac-
ous malignancies. Cutis 2016;97:415-20. the European Dermatology Forum — short tinic keratosis: a follow-up on a Cochrane
8. Glogau RG. The risk of progression to version. J Eur Acad Dermatol Venereol review. Br J Dermatol 2013;169:250-9.
invasive disease. J Am Acad Dermatol 2015;29:2069-79. 26. Kirby JS, Gregory T, Liu G, Leslie DL,
2000;42:23-4. 17. Gupta AK, Paquet M, Villanueva E, Miller JJ. Variation in the cost of manag-
9. Criscione VD, Weinstock MA, Naylor Brintnell W. Interventions for actinic kera- ing actinic keratosis. JAMA Dermatol 2017;
MF, Luque C, Eide MJ, Bingham SF. Ac- toses. Cochrane Database Syst Rev 2012; 153:264-9.
tinic keratoses: natural history and risk of 12:CD004415. 27. Gupta AK, Cooper EA, Feldman SR,
malignant transformation in the Veterans 18. Pocock SJ, Simon R. Sequential treat- Fleischer AB Jr. A survey of office visits for
Affairs Topical Tretinoin Chemopreven- ment assignment with balancing for prog- actinic keratosis as reported by NAMCS,
tion Trial. Cancer 2009;115:2523-30. nostic factors in the controlled clinical 1990-1999: National Ambulatory Medical
10. Frost C, Williams G, Green A. High trial. Biometrics 1975;31:103-15. Care Survey. Cutis 2002;70:Suppl:8-13.
incidence and regression rates of solar 19. Olsen EA, Abernethy ML, Kulp-Shorten 28. Medicijnkosten home page (https://
keratoses in a Queensland community. C, et al. A double-blind, vehicle-controlled www.medicijnkosten.nl).
J Invest Dermatol 2000;115:273-7. study evaluating masoprocol cream in the 29. Rothman KJ, Lash TL, Greenland S.
11. Pomerantz H, Hogan D, Eilers D, et al. treatment of actinic keratoses on the head Modern epidemiology. 3rd ed:Philadel-
Long-term efficacy of topical fluorouracil and neck. J Am Acad Dermatol 1991;24: phia:Lippincott Williams & Wilkins, 2012.
cream, 5%, for treating actinic keratosis: 738-43. Copyright © 2019 Massachusetts Medical Society.
posting presentations from medical meetings online

Online posting of an audio or video recording of an oral presentation at
a medical meeting, with selected slides from the presentation, is not considered
prior publication. Authors should feel free to call or send email to the
Journal’s Editorial Offices if there are any questions about this policy.


Nej Mo A 1811850

Uploaded by

Copyright:

Available Formats

Nej Mo A 1811850

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Nej Mo A 1811850

Uploaded by

Copyright:

Available Formats

The n e w e ng l a n d j o u r na l of m e dic i n e

Randomized Trial of Four Treatment

n engl j med 380;10 nejm.org March 7, 2019 935

936 n engl j med 380;10 nejm.org March 7, 2019

The New England Journal of Medicine

n engl j med 380;10 nejm.org March 7, 2019 937

The New England Journal of Medicine

n engl j med 380;10 nejm.org March 7, 2019

Copyright © 2019 Massachusetts Medical Society. All rights reserved.

1174 Patients were assessed for eligibility

550 Declined to participate

624 Underwent randomization

4 Did not begin treat- 3 Did not begin treat-

14 Had treatment failure 37 Had treatment failure 49 Had treatment failure

Figure 2. Screening, Randomization, and Follow-up.

n engl j med 380;10 nejm.org March 7, 2019 939

940 n engl j med 380;10 nejm.org March 7, 2019

The New England Journal of Medicine

Table 1. Baseline Characteristics of the Patients in the Modified Intention-to-Treat Population.*

n engl j med 380;10 nejm.org March 7, 2019 941

and increase in health-related quality of life were

Detailed results are provided in the Supplemen-

in a continuous area. Findings from the modi-

robustness of the results.

different field-directed and lesion-directed treat-

been published. One meta-analyis, by Vegter and

Tolley,24 indicated that aminolevulinic acid PDT

(ALA-PDT) with the use of BF-200 ALA gel re-

sulted in the highest probability (75.8%) of total

† Cumulative probabilities were based on Kaplan–Meier analysis.

follow-up of 3 to 12 months. But this meta-

analysis did not include 5% fluorouracil, as was

complete clearance of all lesions was assessed.

Forum guidelines, the majority of experts did not

mebutate (0.015% or 0.050%), and 0.5% fluoro-

multiple actinic keratosis lesions.16 However,

5% fluorouracil. In our trial, we used the most

commonly prescribed dosing regimens of the

942 n engl j med 380;10 nejm.org March 7, 2019

The New England Journal of Medicine

n engl j med 380;10 nejm.org March 7, 2019 943

Table 3. Adverse Events.*

944 n engl j med 380;10 nejm.org March 7, 2019

The New England Journal of Medicine

n engl j med 380;10 nejm.org March 7, 2019 945

posting presentations from medical meetings online

946 n engl j med 380;10 nejm.org March 7, 2019

The New England Journal of Medicine

You might also like

† Cumulative probabilities were based on Kaplan–Meier analysis.